Software engineer building agentic products, LLM tooling, and inference systems.
A Bit About Me
- Most of my work has been on production full-stack systems with DRF, TS ecosystem, async workers, WebSockets, relational databases, and React frontends.
- Lately, I've been working on LLM ablations and model editing, along with experiments in long-context efficiency, grounded QA, and agent reliability.
- I like projects that sit between research, tooling, and real-world products.
Current Focus
- Agentic platforms and async backend systems
- Ablations and Abstentions for LLMs
- Inference efficiency and evaluation harness
Technologies I Have Been Working With
- Languages: Python, TypeScript, JavaScript, Go, C++
- App stack: Django REST Framework, TS based backend frameworks, Channels / ASGI, React, Next.js, Vite, Redux Toolkit
- Async and data: Celery, Django-Q, Redis, PostgreSQL, MongoDB, WebSockets, background workers
- LLM stack: PyTorch, JAX, Transformers, Hugging Face, LoRA / QLoRA, bitsandbytes
- Retrieval and evaluation: AWS Vectors, FAISS, Weaviate, BM25, citation-grounded evaluation, benchmark harnesses
- Serving and infra: vLLM, SGLang, Ray Serve, Ollama, Cloudflare Workers, Docker, Linux, AWS
Tooling Snapshot
Technical Interests
- Model internals and steering
- Harness Engineering
- Distributed Systems
- Reinforcement Learning
⭐️ From Muhammad_Aaliyan