Manish Meena
Manish Meena is co-founder and principal engineer at Pruning Labs, the Kolkata-based software engineering studio, and a graduate of IIT Kharagpur and IIT Guwahati. He leads technical architecture across the studio's work: production AI systems and retrieval pipelines, durable event-driven automation, and high-performance web platforms. He also writes the studio's engineering notes.
Manish studied at IIT Kharagpur and IIT Guwahati, and co-founded Pruning Labs on a straightforward premise: most systems fail from accumulated complexity rather than from a missing feature, and the fastest way to a working system is usually to remove things.
Manish owns technical architecture across the studio's engagements: the retrieval pipelines and evaluation harnesses behind the AI systems, the durable workflow design behind the automation work, and the performance budgets behind the platforms - including the decision, made regularly, that a client does not need the thing they came in asking for.
The studio's engineering notes are written by Manish, and exist partly so prospective clients can judge the technical bar before the first call.
What Manish works on
- Production AI systems. Retrieval pipelines with measured recall, deterministic boundaries around model outputs, and evaluation suites that gate every prompt change in CI.
- Durable automation. Idempotent, replayable workflows on Temporal and queues, designed so the second run is a no-op rather than a second charge.
- Edge and performance architecture. Next.js on Cloudflare Workers with enforced Core Web Vitals budgets, server rendering by default, and client JavaScript only where interaction requires it.
Areas of depth
- Retrieval-augmented generation
- pgvector and PostgreSQL vector search
- Evaluation harnesses for LLM systems
- Durable execution and idempotent workflows
- Event-driven architecture
- Next.js and React architecture
- Cloudflare Workers and edge computing
- Core Web Vitals performance engineering
- Go, Python, TypeScript, Rust
Writing
Retrieval recall is the RAG metric that actually decides quality
Most RAG systems fail at retrieval, not generation. How to measure recall@k on your own corpus, what a passing number looks like, and why prompt engineering cannot fix a retrieval problem.
When not to build an AI agent
Multi-agent architectures are oversold. A practical test for whether a workload needs an agent, a single model with tools, or ordinary deterministic code.
Idempotency is the whole job
Automation that cannot safely run twice is a liability. Idempotency keys, durable execution, and decline-code-aware retries - the patterns that keep automated money movement correct.
pgvector or a dedicated vector database
When PostgreSQL with pgvector is the right default for RAG, what actually breaks at scale, and the specific signals that justify moving to a dedicated vector store.
INP is the hard Core Web Vital, and it is a JavaScript problem
Interaction to Next Paint is the metric most sites fail, and image compression will not fix it. Where the long tasks come from, how to break them up, and what an INP budget looks like in practice.
Elsewhere
- Indian Institute of Technology Kharagpur
- Indian Institute of Technology Guwahati
- Kolkata, West Bengal, India
Working with the studio starts with a thirty-minute call. The rest of the team is on the team page.