Current

Working at Surge AI, creating RL environments for long-horizon agentic coding tasks used by frontier labs. Most recently a contributor to the sudo L7 benchmark.

In my free time, working with the SO-ARM101 kit for robotics training with LeRobot.

I think the next biggest value-add after LLMs, which are still confined to computers, is embedding AI in hardware so it can act in the physical world. I am looking for niches that interest me in that direction, from autonomous driving and humanoid robots to medical applications and chip design. Even learning the full training pipeline behind Tesla FSD has been inspiring.

Past projects July 2025 – June 2026

  • LoRA Targeting for Persona SFT

    Fine-tuning · MoE · VLM April 2026 – May 2026

    Gradient-guided LoRA targeting for persona SFT on qwen3.5-35b-a3b; Curated 700,000+ noisy discord messages into high quality multi-turn SFT data; lmms-eval OSS contribution (eval harness used by frontier labs & thousands of users). Accelerated evals by 30x with vLLM. Code

  • Browser use agents

    VLM · Agents Late Feb 2026 – June 2026

    Built two browser use agents with frontier models. Includes a self-built stealth browser harness intended for captcha solving. Supports research tasks: answering questions and validating answers based on multi-document source material on the web. Frequently repurposed for any task where agents need access to websites.

  • Lambda Cloud MCP

    MCP · ML infra · Agents May 2026 – June 2026

    MCP surface for Lambda Cloud — auto-provision GPUs, orchestrate ML environments and training jobs, steer agents by text message. Code

  • Legal Agentic RAG System

    RAG · Agents July 2025 – Nov 2025

    Agentic RAG on official legal pages with various hybrid search implementations, reranking, multi-agent systems. 90% retrieval accuracy on evaluations created from high quality synthetic data. Finetuning on Google cloud platform with torch distributed elastic training. 10,000+ HuggingFace dataset downloads. Code · Training infra

  • Llama 3.1 8B Instruction Tuning

    Fine-tuning Late Jan 2026

    LoRA SFT with +52% IFEval improvement (200→305/834) on ~$10 compute, and multiple quantization experiments. Code

  • Large Language Model Pretraining

    Pretraining · DDP Jan 2026

    Pretrained a personally implemented 450M-param transformer on FineWeb-Edu (10B tokens), 8xA100 gpus using DDP — RoPE, SwiGLU, Flash attention, GQA/MHA, KV-Cache, etc. Code · Checkpoint

  • Transformer from Scratch

    Fundamentals Nov 2025 – Dec 2025

    PyTorch "Attention Is All You Need" implementation with encoder-only, decoder-only, and encoder-decoder architectures. Along with modern extensions (SwiGLU, RMSNorm, RoPE) Code

Technologies

Applied AI
Python, AI Agents, Multi-agent systems, Retrieval augmented generation, Finetuning, Evaluations, Large language models (LLM/SLM), Multi-modal models, Vision language models, Claude Agent SDK, Playwright
Applied ML
PyTorch, HuggingFace, Transformers, Pretraining, Supervised finetuning, LoRA/PEFT, Quantization, Unsloth, Bitsandbytes, Weights & Biases, Torch Elastic, Distributed Data Parallel (DDP)
ML infra
Docker, Google Cloud, Google Compute Engine, Google Kubernetes Engine, Lambda Cloud, vLLM, Vector databases, Hierarchical Navigable Small World (HNSW) tuning, ElasticSearch