# Aakash Subedi

> Aakash Subedi is a software engineer and agentic AI engineer based in Bengaluru, India. He builds AI agents, RAG pipelines and agent harnesses, designs REST APIs, and works on API performance and cost optimisation, across web and mobile.

Aakash Subedi is a software engineer and agentic AI engineer based in Bengaluru, India. Currently Software Engineer at Nihin Media K.K.

## Facts

- Name: Aakash Subedi
- Role: Software Engineer & Agentic AI Engineer
- Location: Bengaluru, India
- Current company: Nihin Media K.K. (January 2026 - Present)
- Focus: Agentic AI, AI agents, Agent harnesses, Retrieval-augmented generation (RAG), Large language models, REST API design, API performance, Cloud cost optimisation, Web development, Mobile development
- Stack (languages): TypeScript, Python, Go
- Stack (ai): AI agents, RAG, agent harnesses, Vercel AI SDK, OpenAI SDK, Ollama, llama.cpp
- Stack (backend): REST APIs, Node.js, Bun, Express, NestJS, FastAPI, BullMQ
- Stack (data): PostgreSQL, Redis, Drizzle
- Stack (web): React, Next.js
- Stack (infra): Docker, AWS
- Education: Bachelor in Computer Science And Engineering (2022-2026)
- Website: https://www.aksh.space
- Email: contact.aakash77@gmail.com

## Experience

### Software Engineer - Nihin Media K.K. (January 2026 - Present)
- building AI agents and RAG pipelines - retrieval, tools and the harness the agent runs in
- designing and shipping REST APIs
- API performance - finding the slow paths and making them faster
- cost optimisation - making things cheaper to run without making them worse

### SDE Intern - Ainocular (July 2025 - December 2025)
- AI-powered platform for intelligent business insights.

### SDE Intern - Lookaround.in (Jan 2025 - June 2025)
- Location discovery platform connecting people to local experiences.

## Projects

- [Qwen PR Review Agent](https://github.com/subediaakash/qwen-backed-cli-coding-assistant) (TypeScript · Bun · 2026): A PR reviewer that runs fully on a local Qwen model through Ollama. I built it to understand how an agent harness really works - the tool loop, guardrails, context and output parsing. Every finding has to quote evidence from the diff, and the harness drops anything it can't verify. There's also an eval runner that plants bugs in a branch and scores recall and false positives.
- [LLM Gateway](https://github.com/subediaakash/llm-gateway) (Go · 2026): llama-server on a port is not an API, so I wrote the layer in front of it. It's in Go and OpenAI-compatible, so any OpenAI SDK just works against my local Qwen3. API keys are stored as SHA-256 digests, rate limits are a token bucket per key, SSE streams tokens straight through, and a model registry decides which backend answers. Ships as a 31 MB container.
- [Express Backend Agent](https://github.com/subediaakash/agent-that-writes-backend) (TypeScript · 2026): You describe a backend, it writes one that compiles. A planner agent makes the file plan, then files are written layer by layer, checked with real tools (prisma validate, tsc) and sent back to the model with the exact errors until they pass. Jobs run through BullMQ on Redis, so the API stays fast while workers do the heavy part.
- [Zomato MCP](https://github.com/subediaakash/zomato-mcp) (TypeScript · 2025): My take on Zomato's MCP idea. A chat agent on the Vercel AI SDK that checks what's available, places orders and lists them, through typed tool calls that hit Postgres via Drizzle. It's scoped to the logged-in user, so it can't see anyone else's orders.
- [CLI Music Player](https://github.com/subediaakash/cli-based-music-player) (JavaScript · 2025): Search a song and it plays, right in the terminal. Audio streams from YouTube through mpv, with playlists, auto-play and next/previous. It even installs mpv for you on setup - brew, apt, choco, whatever your system has.

## Writing

- [How Agents Work at Scale](https://www.aksh.space/blogs/how-agents-work-at-scale) (2026-10-18): What it really takes to run an AI agent for many people - queues, workers, real-tool checks and feedback loops - using my agent that writes backends as the example.
- [Evolution Takes Time](https://www.aksh.space/blogs/evolution-takes-time) (2026-02-01): Exploring the journey of growth and the importance of patience in the face of challenges.
- [My 2025 Session](https://www.aksh.space/blogs/my-2025-session) (2025-12-27): Reflecting on a transformative year - from physical transformation to landing my first stipend.

## Profiles

- [GitHub](https://github.com/subediaakash)
- [LinkedIn](https://www.linkedin.com/in/aakash-subedi-a9986625a/)
- [X / Twitter](https://x.com/void_aakash)