Production AI
Reliable AI workflows built around real data, constraints and failure modes.
AI & DATA ENGINEER
Anirudha Kuchibhotla
I build production AI and data systems that are measurable, reliable, and built for real-world use.
Production AI · LLM Evaluation · RAG / Retrieval · Data Engineering
Built to work after the demo is over.
01 / What I do
Where AI meets data, evaluation and reliability.
Evaluation
Quality, grounding, consistency, reliability and failure analysis.
Retrieval / RAG
Evidence retrieval before model reasoning.
Data Engineering
Pipelines, modeling and dependable data foundations.
Trust & Safety
Turning risky system behavior into testable engineering requirements.
02 / Proof of work
AI engineering case study / 2026
DevSignal AI
I built DevSignal to answer a simple question: Can an AI/data system improve in ways we can actually measure?
DevSignal turns historical GitHub issues into searchable engineering knowledge, with evaluation built into the system from the start.
Python / FastAPI / Retrieval / Evaluation / GitHub API
- 15K+
- issues analyzed
- 0.48 → 0.61
- retrieval Recall@5
- 0.13 → 0.61
- label micro-F1
- BM25
- 0.484
- Dense
- 0.581
- HybridSelected approach
- 0.613
Dense retrieval recovered differently worded issues. Lexical search still mattered for exact technical identifiers. The final approach combines both.
03 / Experience
Engineering across production AI and data systems.
2025 - NOW
AAA
AI & Data Engineering
Production AI · Evaluation · Retrieval · Async Processing · Data Systems
Previous
AdaIQ
Data Engineering and Product Analytics
Data Engineering · Product Analytics
Academic
Northeastern
Data Analytics Engineering
Data Analytics · Engineering Systems
Earlier
Cognizant
Enterprise Engineering and Data Systems
Enterprise Engineering · Data Systems
04 / Engineering principles
Simple rules for complicated systems.
- 01Retrieve before reasoning.
- 02Evaluate before claiming improvement.
- 03Complexity should earn its place.
- 04Turn production failures into future tests.
05 / About
I'm Anirudha.
I'm interested in the engineering that starts once an AI prototype touches real users, real data and real constraints.
That means thinking about retrieval, evaluation, state, failure handling, data quality and what the system should do when evidence is insufficient.
This has pulled my work toward production AI, evaluation, retrieval and the data systems underneath them.
Read about Anirudha- Domain knowledge
- Requirements
- Data / Rules
- AI System
- Evaluation
- Feedback
- Improvement