I didn't get into technology because of shiny venture capital headlines or superficial hype cycles. Growing up in Kanpur, my obsession started with a simple, persistent reality: the world is drowning in numbers, but starving for clarity. My question was never just about code execution — it was: "What is the data actually trying to tell us, and why are so many people misreading it?"

When I enrolled in B.Tech Artificial Intelligence & Machine Learning at IIMT University (Class of 2027), I avoided the trap of surface-level model building or routine coding drills. Instead, I immersed myself in the core foundations of Data Science: exploratory data analysis, relational SQL modeling, and hands-on research into modern data platforms. I realized that crunching numbers is only half the battle — my true passion lies in data storytelling: transforming complex, messy data into compelling, actionable narratives that drive real-world impact.

In early 2024, everyone was building generic chat wrappers. But if you've ever hooked an LLM directly to a production SQL database, you know the terror: hallucinations, dropped tables, syntax errors, and zero safety guarantees.

I decided to solve this head-on. I architected SQL-GPT-Omni — not as a single prompt, but as a deterministic tri-agent containment pipeline. Step 1 interprets natural language. Step 2 parses and writes dialect-specific SQL. Step 3 intercepts the query through an Abstract Syntax Tree (AST) validator to verify read-only constraints. Step 4 generates plain-English execution explanations. The result was a 92% drop in syntax errors and zero destructive queries.

Next came Promptly Evaluate v2. When working with enterprise AI, you cannot evaluate models on "vibes". I built an automated multi-provider benchmarking harness that quantitatively evaluates prompt outputs for latency, token economics, and hallucination containment.

Engineering isn't tested in a quiet room with endless time. It is tested under bright lights and tight deadlines when everything that can go wrong is waiting to happen.

As Co-Organizer for TEDxIIMTUniversity (Theme: "Break The Flow"), I managed cross-functional operations, speaker hospitality, technical rehearsals, and stage timing for national thought leaders. We maintained zero operational downtime throughout the entire conference. (Verified on TED.com).

Simultaneously, I was chosen for two consecutive editions (2023 & 2024/2025) as Team Lead for Smart India Hackathon (SIH). Leading a 6-person engineering team through 36 sleepless hours — debugging data models at 4:00 AM, keeping morale high, and delivering a functional civic software prototype directly to government juries — forged my leadership under pressure.

In GirlScript Summer of Code (GSSoC), I evaluated and cleaned 100,000+ noisy data records with Pandas/NumPy, built Power BI visual dashboards, and onboarded 50+ new contributors to open-source Git workflows.

This story is only just getting started. I graduate in August 2027, and my immediate mission is to join an ambitious engineering team as a Summer 2026 / 2027 AI/ML Engineering or Data Science Intern.

I bring a rare blend: deep hands-on technical skills in building autonomous agentic loops and production data pipelines, coupled with the tested empathy and communication of someone who has mentored hundreds of peers and commanded high-stakes stages.