# Anas Khan - Software Engineer

> AI-readable profile. For full context: https://anaskhan.me

## Bio

Software engineer working on backend systems and applied AI. I build customer-facing data engineering and agentic developer experiences across Microsoft Fabric. During Google Summer of Code 2025 at Google DeepMind, I worked on LLM evaluation pipelines and multimodal benchmarks, and I previously worked as an SDE at HackerRank.

My engineering spans agentic AI systems, LLM evaluation frameworks, benchmark design, and scalable backend infrastructure.

## Focus Areas

- **LLM Evaluation & Benchmarking**: Designing evaluation harnesses, automated scoring pipelines, and extended benchmark suites for frontier and open-source models
- **Agentic AI**: Autonomous agent systems, tool-use pipelines, multi-agent orchestration, and AI-assisted software engineering
- **Applied AI Engineering**: Production LLM integration, RAG architectures, fine-tuning (LoRA/QLoRA), efficient inference
- **Backend Systems**: High-throughput APIs, distributed systems, data pipelines, and scalable infrastructure

## Technical Skills

### AI / LLMs

- **Evaluation**: LM-Eval Harness, HELM, custom eval pipelines, automated benchmarking, iterative scoring
- **Frameworks**: HuggingFace Transformers, LangChain, LlamaIndex, vLLM, Ollama
- **Models evaluated**: Gemini, Gemma, LLaMA 3.x, DeepSeek, Phi-4, Mistral, GPT-4
- **Techniques**: RAG, fine-tuning (LoRA, QLoRA), prompt engineering, tool-use, function calling, RLHF
- **Agentic**: ReAct agents, multi-agent systems, autonomous coding agents, agentic workflows
- **Multimodal**: Vision-language model evaluation, image-to-code benchmarks, UI understanding

### Languages

- Python (primary), Go, TypeScript, JavaScript, Ruby, SQL

### Frameworks & Libraries

- React, Django, Flask, FastAPI, Rails, ExpressJS, NodeJS

### Infrastructure

- AWS (EC2, EKS, Lambda), GCP (BigQuery, GCS, Cloud Run), Docker, Kubernetes, Celery, Airflow, dbt

### Databases

- PostgreSQL, MongoDB, SQLite, MariaDB, Redis, BigQuery

## Experience

### Microsoft - Software Engineer

**May 2026 - Present** | TypeScript, React, LLMs, Spark

Building customer-facing data engineering and agentic developer experiences across Microsoft Fabric, including Lakehouse, Notebooks, Spark Job Definitions, Synapse Studio, and monitoring, within Microsoft Cloud + AI (Org).

### CircuitVerse - Google Summer of Code '26 Mentor

**May 2026 - Present** | Ruby, Rails, Vue

Mentoring the [Assignment Suite & Classroom Workflow Enhancement](https://summerofcode.withgoogle.com/organizations/circuitverseorg/projects/details/JI2PN30m) project for GSoC 2026. Guiding the implementation of structured subgroup team management, collaborative group assignment workflows, circuit template libraries with simulation-based auto-grading test cases, and a complete LTI 1.3 LMS integration (OIDC, OAuth 2.0, AGS, NRPS).

### HackerRank - Software Development Engineer

**Jan 2025 - May 2026** | Ruby, Rails, Python, LLMs

Worked full-time as an SDE across Payments, LLMs, Voice, and new products.

- Worked on [HackerRank MCP](https://support.hackerrank.com/articles/8142080826-july-2026-release-notes#hackerrank-mcp-limited-availability)
- Worked on [HackerRank Orchestrate](https://www.hackerrank.com/hackerrank-orchestrate-may26), including the evaluation pipeline, Agent Files, and voice-based AI judge interviews for its AI agent hackathons. Read [Behind the Scenes of HackerRank Orchestrate](https://www.hackerrank.com/blog/behind-the-scenes-of-hackerrank-orchestrate/) (ghostwritten for CEO)
- Worked on [AI-powered Mock Interviews](https://www.hackerrank.com/dashboard) for recruiter technical screens, behavioral rounds, AI fluency, and more
- Contributed to HackerRank's open source [Hiring Agent](https://github.com/interviewstreet/hiring-agent), a resume evaluation and scoring pipeline
- Built [HackerRank Learn](https://www.hackerrank.com/learn), an early proof of concept for hands-on AI engineering courses built around real projects

Previously as SDE Intern on the HackerRank Labs team:

- Benchmarked and fine-tuned open-source LLMs (LLaMA 3.2, DeepSeek-7B, Gemma 3, Phi-4 14B) for efficient inference and domain-specific task adaptation
- Built a scalable distributed web crawler and scraper with Python, Playwright, and NLTK achieving 94% automated labeling accuracy
- Developed a Slackbot with Ruby on Rails and Salesforce API to automate enterprise customer onboarding workflows
- Worked directly under Hari (CTO) as manager

### Google DeepMind - Google Summer of Code '25 Contributor

**May 2025 - Nov 2025** | Python, TypeScript, LLMs, Evaluation, Research

- Designed and ran evaluation pipelines for Gemini, Gemma, and other MLLMs on extended open-source coding benchmarks
- Focused on multimodal capabilities: visual understanding, frontend code generation from designs, and UI reasoning
- Built iterative evaluation infrastructure with automated result publishing to Hugging Face Spaces
- Extended benchmark suites to cover multi-step reasoning, image-to-code fidelity, and component-level accuracy

### FamPay (YC S19) - Backend Engineer Intern

**Aug 2024 - Nov 2024** | Python, Django, Go, Postgres, AWS, Celery, Docker

- Developed a scalable data deletion service for regulatory (DPDP) compliance
- Built recharge and payment use cases on TPAP infrastructure using hexagonal architecture in Go
- Enabled in-app P2M (Pay-to-Merchant) transactions supporting 100,000+ users

### General Mills - Digital & Technology Intern

**June 2024 - Aug 2024** | Python, SQL, Airflow, dbt, GCP, BigQuery

- Worked on commercial data foundation (File Data Ingestion) project with BigQuery
- Standardized incoming file management via shared mailbox, SFTP, and GCP Buckets
- Built data quality checks using DBT integrated into Airflow DAGs with Python and SQL

### Frappe Technologies - SDE Intern

**Feb 2024 - Apr 2024 | June 2023 - Aug 2023** | Python, JavaScript, MariaDB, Redis

- Contributed to open-source Frappe Framework for flagship ERPNext product
- Engineered automated deployment tooling achieving 47% reduction in setup time
- Architected an inventory module app leveraging DocTypes for stock tracking and valuation

### MohiteTax - Python Developer Intern

**April 2023 - June 2023** | Python, Selenium, MongoDB, AWS

- Automated data collection for 2.7M+ company records via MCA Services using Selenium
- Deployed on AWS infrastructure (EC2, EKS, Lambda) achieving 65% reduction in scraping time

## Projects

### CodeClip

Code snippet sharing platform with customized URLs, password protection, QR code generation, and a user dashboard. Built with Flask and the CS50 library as a first full-stack milestone.
→ https://github.com/anxkhn/codeclip

## Recent contributions

I like building products and making them better. A lot of that happens in the open. Right now that is 503 contributions across 90 repositories, mostly on projects I care about or can usefully improve.

### Open source work by category

Sorted by stars within each category. Links filter to my PRs.

#### AI / Agents / LLMs

- [ollama/ollama](https://github.com/ollama/ollama/pulls?q=is:pr+is:merged+author:anxkhn) — run open-source and frontier models locally; 2 PRs · 180k stars
- [unslothai/unsloth](https://github.com/unslothai/unsloth/pulls?q=is:pr+is:merged+author:anxkhn) — local UI to run and train LLMs and diffusion models; 7 PRs · 76k stars
- [docling-project/docling](https://github.com/docling-project/docling/pulls?q=is:pr+is:merged+author:anxkhn) — document parsing pipeline for gen AI; 14 PRs · 66k stars
- [mem0ai/mem0](https://github.com/mem0ai/mem0/pulls?q=is:pr+is:merged+author:anxkhn) — universal memory layer for AI agents; 1 PR · 65k stars
- [BerriAI/litellm](https://github.com/BerriAI/litellm/pulls?q=is:pr+is:merged+author:anxkhn) — unified API to call 100+ LLM providers; 1 PR · 58k stars
- [ComposioHQ/composio](https://github.com/ComposioHQ/composio/pulls?q=is:pr+is:merged+author:anxkhn) — toolkits, auth, and workbench for AI agents that take action; 6 PRs · 30k stars
- [openai/openai-agents-python](https://github.com/openai/openai-agents-python/pulls?q=is:pr+is:merged+author:anxkhn) — OpenAI's Python framework for multi-agent workflows; 5 PRs · 29k stars
- [deepset-ai/haystack](https://github.com/deepset-ai/haystack/pulls?q=is:pr+is:merged+author:anxkhn) — AI orchestration framework for RAG, agents, and LLM apps; 1 PR · 26k stars
- [promptfoo/promptfoo](https://github.com/promptfoo/promptfoo/pulls?q=is:pr+is:merged+author:anxkhn) — OpenAI's LLM eval, red-teaming, and prompt testing toolkit; 5 PRs · 25k stars
- [comet-ml/opik](https://github.com/comet-ml/opik/pulls?q=is:pr+is:merged+author:anxkhn) — debug, evaluate, and monitor LLM apps and agent workflows; 1 PR · 22k stars
- [google/adk-python](https://github.com/google/adk-python/commits?author=anxkhn) — Google's Agent Development Kit for building and evaluating agents; 17 commits · 21k stars
- [SWE-agent/SWE-agent](https://github.com/SWE-agent/SWE-agent/pulls?q=is:pr+is:merged+author:anxkhn) — autonomous agent that fixes GitHub issues with your LM; 6 PRs · 20k stars
- [googleapis/mcp-toolbox](https://github.com/googleapis/mcp-toolbox/pulls?q=is:pr+is:merged+author:anxkhn) — Google's open-source MCP server for databases; 9 PRs · 16k stars
- [pipecat-ai/pipecat](https://github.com/pipecat-ai/pipecat/pulls?q=is:pr+is:merged+author:anxkhn) — open-source framework for voice agents and realtime AI; 6 PRs · 15k stars
- [ollama/ollama-python](https://github.com/ollama/ollama-python/pulls?q=is:pr+is:merged+author:anxkhn) — official Python library for Ollama; 2 PRs · 10k stars
- [omnigent-ai/omnigent](https://github.com/omnigent-ai/omnigent/pulls?q=is:pr+is:merged+author:anxkhn) — open-source AI agent framework and meta-harness; 22 PRs · 10k stars
- [NVIDIA/garak](https://github.com/NVIDIA/garak/pulls?q=is:pr+is:merged+author:anxkhn) — NVIDIA's LLM vulnerability scanner; 7 PRs · 9k stars
- [k8sgpt-ai/k8sgpt](https://github.com/k8sgpt-ai/k8sgpt/pulls?q=is:pr+is:merged+author:anxkhn) — AI-powered Kubernetes troubleshooting CLI; 19 PRs · 8k stars
- [kagent-dev/kagent](https://github.com/kagent-dev/kagent/pulls?q=is:pr+is:merged+author:anxkhn) — cloud-native agentic AI for Kubernetes; 10 PRs · 4k stars
- [agent-of-empires/agent-of-empires](https://github.com/agent-of-empires/agent-of-empires/pulls?q=is:pr+is:merged+author:anxkhn) — manage multiple coding agents from TUI or web; 1 PR · 3k stars
- [coddingtonbear/obsidian-local-rest-api](https://github.com/coddingtonbear/obsidian-local-rest-api/pulls?q=is:pr+is:merged+author:anxkhn) — a secure REST API and Model Context Protocol (MCP) server for your vault.; 1 PR · 3k stars
- [UKGovernmentBEIS/inspect_ai](https://github.com/UKGovernmentBEIS/inspect_ai/pulls?q=is:pr+is:merged+author:anxkhn) — framework for large language model evaluations; 5 PRs · 3k stars
- [org2AI/ORG2](https://github.com/org2AI/ORG2/pulls?q=is:pr+is:merged+author:anxkhn) — system of record for how agents build software; 1 PR · 3k stars
- [yctimlin/mcp_excalidraw](https://github.com/yctimlin/mcp_excalidraw/pulls?q=is:pr+is:merged+author:anxkhn) — MCP server for creating and editing Excalidraw diagrams; 1 PR · 2k stars
- [a2aproject/a2a-python](https://github.com/a2aproject/a2a-python/pulls?q=is:pr+is:merged+author:anxkhn) — official Python SDK for the Agent2Agent (A2A) Protocol; 1 PR · 2k stars
- [google-deepmind/mujoco_warp](https://github.com/google-deepmind/mujoco_warp/pulls?q=is:pr+is:merged+author:anxkhn) — GPU-optimized MuJoCo physics simulator from DeepMind; 2 PRs · 1k stars
- [sipyourdrink-ltd/bernstein](https://github.com/sipyourdrink-ltd/bernstein/pulls?q=is:pr+is:merged+author:anxkhn) — deterministic orchestrator for CLI coding agents; 1 PR · 1k stars
- [johannesjo/parallel-code](https://github.com/johannesjo/parallel-code/pulls?q=is:pr+is:merged+author:anxkhn) — run Claude Code, Codex, and Gemini in parallel worktrees; 1 PR · 1k stars
- [dapr/dapr-agents](https://github.com/dapr/dapr-agents/pulls?q=is:pr+is:merged+author:anxkhn) — autonomous AI agents with Dapr workflow and telemetry; 11 PRs · 744 stars
- [a2aproject/a2a-js](https://github.com/a2aproject/a2a-js/pulls?q=is:pr+is:merged+author:anxkhn) — official JavaScript SDK for the Agent2Agent (A2A) Protocol; 4 PRs · 609 stars
- [google-deepmind/simply](https://github.com/google-deepmind/simply/pulls?q=is:pr+is:merged+author:anxkhn) — minimal JAX research codebase for LLMs; 1 PR · 569 stars
- [mco-org/mco](https://github.com/mco-org/mco/pulls?q=is:pr+is:merged+author:anxkhn) — CLI orchestration for AI coding agents in parallel; 1 PR · 514 stars
- [llm-d/llm-d-router](https://github.com/llm-d/llm-d-router/pulls?q=is:pr+is:merged+author:anxkhn) — intelligent router entry point for LLM inference requests; 9 PRs · 330 stars

#### Developer tools / frameworks

- [react/react](https://github.com/react/react/pulls?q=is:pr+is:merged+author:anxkhn) — library for web and native user interfaces; 4 PRs · 250k stars
- [torvalds/linux](https://github.com/torvalds/linux/commits?author=anxkhn&since=2026-01-01) — Linux kernel source tree; 7 commits · 248k stars
- [astral-sh/uv](https://github.com/astral-sh/uv/pulls?q=is:pr+is:merged+author:anxkhn) — extremely fast Python package and project manager; 2 PRs · 90k stars
- [rails/rails](https://github.com/rails/rails/pulls?q=is:pr+is:merged+author:anxkhn) — full-stack web framework for Ruby; 2 PRs · 59k stars
- [reflex-dev/reflex](https://github.com/reflex-dev/reflex/pulls?q=is:pr+is:merged+author:anxkhn) — build web apps in pure Python; 11 PRs · 29k stars
- [kucherenko/jscpd](https://github.com/kucherenko/jscpd/pulls?q=is:pr+is:merged+author:anxkhn) — copy/paste detector for source code; 1 PR · 6k stars

#### Observability

- [prometheus/prometheus](https://github.com/prometheus/prometheus/pulls?q=is:pr+is:merged+author:anxkhn) — monitoring system and time series database; 4 PRs · 66k stars
- [thanos-io/thanos](https://github.com/thanos-io/thanos/pulls?q=is:pr+is:merged+author:anxkhn) — highly available Prometheus with long-term storage; 1 PR · 14k stars
- [cortexproject/cortex](https://github.com/cortexproject/cortex/pulls?q=is:pr+is:merged+author:anxkhn) — horizontally scalable multi-tenant Prometheus; 1 PR · 6k stars
- [perses/perses](https://github.com/perses/perses/pulls?q=is:pr+is:merged+author:anxkhn) — CNCF sandbox for observability visualization; 24 PRs · 2k stars

#### Scraping / browser / sandbox

- [cypress-io/cypress](https://github.com/cypress-io/cypress/pulls?q=is:pr+is:merged+author:anxkhn) — end-to-end testing for anything that runs in a browser; 11 PRs · 51k stars
- [apify/crawlee](https://github.com/apify/crawlee/pulls?q=is:pr+is:merged+author:anxkhn) — web scraping and browser automation for Node.js; 13 PRs · 26k stars
- [Skyvern-AI/skyvern](https://github.com/Skyvern-AI/skyvern/pulls?q=is:pr+is:merged+author:anxkhn) — AI browser agent for web workflows; 2 PRs · 23k stars
- [e2b-dev/E2B](https://github.com/e2b-dev/E2B/pulls?q=is:pr+is:merged+author:anxkhn) — open-source secure sandboxes for enterprise-grade agents; 3 PRs · 14k stars
- [apify/crawlee-python](https://github.com/apify/crawlee-python/pulls?q=is:pr+is:merged+author:anxkhn) — web scraping and browser automation for Python; 24 PRs · 9k stars
- [dagger/container-use](https://github.com/dagger/container-use/pulls?q=is:pr+is:merged+author:anxkhn) — isolated container environments for coding agents; 1 PR · 4k stars

#### Data engineering

- [apache/airflow](https://github.com/apache/airflow/pulls?q=is:pr+is:merged+author:anxkhn) — platform to author, schedule, and monitor workflows; 3 PRs · 47k stars
- [apache/arrow](https://github.com/apache/arrow/pulls?q=is:pr+is:merged+author:anxkhn) — columnar in-memory format for fast analytical data; 1 PR · 17k stars
- [fivetran/great_expectations](https://github.com/fivetran/great_expectations/pulls?q=is:pr+is:merged+author:anxkhn) — data quality and validation framework; 3 PRs · 12k stars
- [sqlfluff/sqlfluff](https://github.com/sqlfluff/sqlfluff/pulls?q=is:pr+is:merged+author:anxkhn) — modular SQL linter and auto-formatter; 15 PRs · 10k stars
- [apache/iceberg](https://github.com/apache/iceberg/pulls?q=is:pr+is:merged+author:anxkhn) — open table format for large analytic datasets; 4 PRs · 9k stars
- [dlt-hub/dlt](https://github.com/dlt-hub/dlt/pulls?q=is:pr+is:merged+author:anxkhn) — Python data load tool (dlt) for easy data loading; 3 PRs · 6k stars
- [apache/calcite](https://github.com/apache/calcite/pulls?q=is:pr+is:merged+author:anxkhn) — dynamic data management framework; 3 PRs · 5k stars
- [unitycatalog/unitycatalog](https://github.com/unitycatalog/unitycatalog/pulls?q=is:pr+is:merged+author:anxkhn) — open multi-modal catalog for data and AI; 3 PRs · 4k stars
- [SQLMesh/sqlmesh](https://github.com/SQLMesh/sqlmesh/pulls?q=is:pr+is:merged+author:anxkhn) — scalable data transformation framework, dbt-compatible; 9 PRs · 3k stars
- [apache/parquet-java](https://github.com/apache/parquet-java/pulls?q=is:pr+is:merged+author:anxkhn) — Apache Parquet columnar format for Java; 3 PRs · 3k stars
- [apache/polaris](https://github.com/apache/polaris/pulls?q=is:pr+is:merged+author:anxkhn) — open catalog for Apache Iceberg; 6 PRs · 2k stars
- [openGemini/openGemini](https://github.com/openGemini/openGemini/pulls?q=is:pr+is:merged+author:anxkhn) — CNCF distributed time-series database; 9 PRs · 1k stars
- [apache/iceberg-python](https://github.com/apache/iceberg-python/pulls?q=is:pr+is:merged+author:anxkhn) — PyIceberg, Python library for Apache Iceberg tables; 13 PRs · 1k stars

#### Cloud native / Kubernetes

- [kestra-io/kestra](https://github.com/kestra-io/kestra/pulls?q=is:pr+is:merged+author:anxkhn) — event-driven orchestration and scheduling platform; 14 PRs · 28k stars
- [dapr/dapr](https://github.com/dapr/dapr/pulls?q=is:pr+is:merged+author:anxkhn) — portable runtime for distributed cloud and edge apps; 7 PRs · 26k stars
- [rook/rook](https://github.com/rook/rook/pulls?q=is:pr+is:merged+author:anxkhn) — storage orchestration for Kubernetes; 12 PRs · 14k stars
- [higress-group/higress](https://github.com/higress-group/higress/pulls?q=is:pr+is:merged+author:anxkhn) — AI-native API gateway; 1 PR · 9k stars
- [tektoncd/pipeline](https://github.com/tektoncd/pipeline/pulls?q=is:pr+is:merged+author:anxkhn) — cloud-native CI/CD pipeline resource for Kubernetes; 7 PRs · 9k stars
- [kubevela/kubevela](https://github.com/kubevela/kubevela/pulls?q=is:pr+is:merged+author:anxkhn) — modern cloud-native application platform; 6 PRs · 8k stars
- [flyteorg/flyte](https://github.com/flyteorg/flyte/pulls?q=is:pr+is:merged+author:anxkhn) — dynamic AI workflow orchestration for data and models; 1 PR · 7k stars
- [cloud-custodian/cloud-custodian](https://github.com/cloud-custodian/cloud-custodian/pulls?q=is:pr+is:merged+author:anxkhn) — rules engine for cloud security, cost, and governance; 10 PRs · 6k stars
- [volcano-sh/volcano](https://github.com/volcano-sh/volcano/pulls?q=is:pr+is:merged+author:anxkhn) — cloud-native batch system under CNCF; 2 PRs · 6k stars
- [strimzi/strimzi-kafka-operator](https://github.com/strimzi/strimzi-kafka-operator/pulls?q=is:pr+is:merged+author:anxkhn) — Apache Kafka on Kubernetes; 3 PRs · 6k stars
- [kserve/kserve](https://github.com/kserve/kserve/pulls?q=is:pr+is:merged+author:anxkhn) — standardized AI inference platform on Kubernetes; 1 PR · 6k stars
- [kgateway-dev/kgateway](https://github.com/kgateway-dev/kgateway/pulls?q=is:pr+is:merged+author:anxkhn) — cloud-native API gateway and AI gateway; 1 PR · 6k stars
- [dragonflyoss/dragonfly](https://github.com/dragonflyoss/dragonfly/pulls?q=is:pr+is:merged+author:anxkhn) — P2P data distribution and image acceleration; 5 PRs · 3k stars
- [open-feature/flagd](https://github.com/open-feature/flagd/pulls?q=is:pr+is:merged+author:anxkhn) — feature flag daemon with a Unix philosophy; 10 PRs · 989 stars
- [openebs/dynamic-localpv-provisioner](https://github.com/openebs/dynamic-localpv-provisioner/pulls?q=is:pr+is:merged+author:anxkhn) — dynamic node-local persistent volumes for Kubernetes; 8 PRs · 211 stars
- [kubeflow/sdk](https://github.com/kubeflow/sdk/pulls?q=is:pr+is:merged+author:anxkhn) — Python SDK to run AI workloads on Kubernetes; 3 PRs · 147 stars

#### Education / community

- [TheAlgorithms/C](https://github.com/TheAlgorithms/C/pulls?q=is:pr+is:merged+author:anxkhn) — educational algorithms and data structures in C; 1 PR · 22k stars
- [fastapi/sqlmodel](https://github.com/fastapi/sqlmodel/pulls?q=is:pr+is:merged+author:anxkhn) — sQL databases in Python, designed for simplicity, compatibility, and robustness.; 1 PR · 18k stars
- [open-telemetry/opentelemetry-collector](https://github.com/open-telemetry/opentelemetry-collector/pulls?q=is:pr+is:merged+author:anxkhn) — openTelemetry Collector; 1 PR · 8k stars
- [chaitanyagiri/munder-difflin](https://github.com/chaitanyagiri/munder-difflin/pulls?q=is:pr+is:merged+author:anxkhn) — local multi-agent harness; 1 PR · 7k stars
- [google-deepmind/open_spiel](https://github.com/google-deepmind/open_spiel/pulls?q=is:pr+is:merged+author:anxkhn) — openSpiel is a collection of environments and algorithms for research in general…; 2 PRs · 5k stars
- [CircuitVerse/CircuitVerse](https://github.com/CircuitVerse/CircuitVerse/pulls?q=is:pr+is:merged+author:anxkhn) — online digital logic circuit simulator; 2 PRs · 1k stars
- [code100x/daily-code](https://github.com/code100x/daily-code/pulls?q=is:pr+is:merged+author:anxkhn) — daily coding practice platform; 2 PRs · 956 stars
- [0xku/leetcode-compensation](https://github.com/0xku/leetcode-compensation/pulls?q=is:pr+is:merged+author:anxkhn) — India software engineering salaries from LeetCode compensation; 4 PRs · 415 stars
- [codinasion/codinasion](https://github.com/codinasion/codinasion/pulls?q=is:pr+is:merged+author:anxkhn) — open-source collaboration and learning projects; 4 PRs · 318 stars
- [fossunited/fossunited](https://github.com/fossunited/fossunited/pulls?q=is:pr+is:merged+author:anxkhn) — FOSS United community platform; 1 PR · 211 stars
- [google-deepmind/jax_privacy](https://github.com/google-deepmind/jax_privacy/pulls?q=is:pr+is:merged+author:anxkhn) — algorithms for Privacy-Preserving Machine Learning in JAX; 1 PR · 193 stars
- [fayazara/Kaze](https://github.com/fayazara/Kaze/pulls?q=is:pr+is:merged+author:anxkhn) — local speech-to-text app; 1 PR · 174 stars

## Current Interests

- Agentic AI systems and autonomous software engineering agents
- LLM evaluation methodology, benchmark design, and leaderboard infrastructure
- Multimodal model capabilities for code and UI understanding
- Efficient inference (quantization, speculative decoding, batching strategies)
- Open-source model fine-tuning for domain-specific tasks
- Building developer tools powered by LLMs

## Writing

Technical blog covering AI, backend engineering, LLMs, and dev tooling.
→ https://blog.anaskhan.me

- [How I booked my IndiaFOSS ticket at the thirteenth hour](https://blog.anaskhan.me/indiafoss-ticket) — Sep 2026
- [REST, GraphQL, or gRPC: Choosing the Right API Paradigm](https://blog.anaskhan.me/api-paradigms) — Oct 2025
- [Decoding Interview Coder](https://blog.anaskhan.me/decoding-interview-coder) — Apr 2025
- [My Atlan Interview Experience](https://blog.anaskhan.me/atlan-interview-experience) — Dec 2024
- [General Mills Interview Experience](https://blog.anaskhan.me/gm-interview) — Sep 2024

## Contact

- **Email**: anxkhn@duck.com
- **Website**: https://anaskhan.me
- **GitHub**: https://github.com/anxkhn
- **LinkedIn**: https://linkedin.com/in/anxkhn
- **Twitter/X**: https://x.com/anxkhn
- **Resume**: https://anaskhan.me/cv
- **Schedule a call**: https://cal.com/anxkhn/catchup