The most dangerous assumption in AI is that the model is the product. It isn't.
The system around it — the data pipelines, the evaluation harness, the deployment strategy — that's the product. The model is just a component.
I build AI systems that are actually deployed — not just impressive in notebooks. My work spans LLM application development, computer vision pipelines, and the infrastructure that keeps both running in production.
“Engineering intelligence means understanding where models fail, and building for the humans on the other end.”
— Jegan.T, AI Engineer
CURRENT FOCUS
Production LLM systems with agentic reasoning.
AVAILABILITY
Open to full-time and contract roles.
LOCATION
Remote-first. Based in India.
MAR 2025 · 4 MIN READ
Most retrieval-augmented generation systems fail not because the embeddings are wrong, but because the chunking strategy misunderstands how humans ask questions.
The distance between the query and the relevant passage isn't semantic — it's structural. When you optimize for retrieval accuracy in isolation, you often destroy the contextual coherence that makes an answer actually useful. Fix the chunking before you tune the embeddings.
FEB 2025 · 6 MIN READ
A model with 94% accuracy and one with 91% accuracy can produce identical user experiences.
What differentiates them isn't the delta in benchmark performance — it's how they handle the 6% and 9% of cases where they fail. Graceful degradation, uncertainty communication, and fallback design are the real product decisions. Benchmark chasing is a distraction from the work that actually ships.
JAN 2025 · 5 MIN READ
Computer vision models are remarkably capable at pattern recognition in controlled distributions. They remain fundamentally brittle everywhere else.
The engineering challenge isn't building a model that works — it's building a system that knows when it doesn't work, and communicates that uncertainty in a way humans can act on. Domain shift, lighting variation, adversarial inputs: each one is an argument for better system design, not better architecture.
LANGUAGE
6 TOOLS TRACKED
VISION
6 TOOLS TRACKED
OPERATIONS
6 TOOLS TRACKED
ANALYTICS
6 TOOLS TRACKED
JEGAN.T
AI ENGINEER
SUBJECT demonstrates advanced capability in designing and deploying AI systems under production constraints.
Not purely a model builder — SUBJECT operates across the full stack: data, model, system, and interface. Has a documented history of reducing model deployment timelines, improving inference efficiency, and building evaluation frameworks that catch regressions before they reach users.
Prefers uncomfortable questions about data quality over comfortable conversations about model architecture. Got into AI not because of the hype — but because understanding how humans process information is, still, the most interesting engineering problem there is.
2024 — PRESENT
Senior AI Engineer
Company / Freelance
Leading production LLM deployments and computer vision integrations for enterprise clients.
2022 — 2024
Machine Learning Engineer
Previous Company
Built and maintained ML infrastructure handling 2M+ daily predictions. Led team of 4.
2020 — 2022
Data Scientist
Earlier Role
Developed forecasting models and deployed first production ML pipelines.
2016 — 2020
B.Tech Computer Science
University Name
Graduated with honors. Thesis: Deep learning for medical imaging.
Open to full-time roles, contract engagements, and interesting problems. If you're building something that requires an engineer who thinks about the whole system — let's talk.
TRANSMIT →● SIGNAL ACTIVE · RESPONSE < 48HRS