_{B.Tech CSE @ PES University · Bengaluru}

AI / ML ENGINEER   ·   SYSTEMS BUILDER   ·   AGENTIC AI   ·   AI SAFETY

I build intelligent systems across machine learning, agentic AI, backend engineering and AI safety research.

My focus is not only whether a model produces the right output — but what happens underneath it: representations, internal dynamics, orchestration, reliability, observability and safety.

Build the system. Instrument the system. Understand the system.

NOW BUILDING

EDGE DAEMON

AI SYSTEMS · MECHANISTIC MONITORING · SAFETY

EdgeDaemon is my current research / venture direction focused on infrastructure for observing intelligent systems below the conventional input/output layer.

MECHGUARD

MechGuard is the flagship system — a mechanistic AI safety monitoring framework spanning training-time and deployment-time signals.

ATTEST Training-time Investigating geometric signals during fine-tuning and whether internal changes can reveal emergent behavioral shifts. geometry · fine-tuning · phase transitions

WATCH Deployment-time Investigating activation-level signatures associated with coordination / collusion in multi-agent systems. activations · representations · multi-agent

TRAINING → INTERNAL DYNAMICS → BEHAVIORAL TRANSITION → DEPLOYMENT → INTERNAL REPRESENTATIONS → SAFETY

SELECTED WORK

systems · agents · ML · applied intelligence

⚙️ ML PIPELINE ORCHESTRATOR Autonomous ML workflow orchestration across data preparation, experimentation and evaluation. Python · ML · Agents · Automation ↗ VIEW PROJECT	🎙️ SUNRISE AI Voice-first conversational AI combining speech interaction, reasoning and tool-oriented behavior. Python · LLMs · Voice AI · Agents ↗ VIEW PROJECT
🌾 MILLETDATANET Applied agricultural ML connecting data processing with predictive workflows. Python · Pandas · scikit-learn ↗ VIEW PROJECT	🌦️ WEATHER RISK Weather-risk forecasting pipeline paired with an application layer. Python · ML · Forecasting MODEL · APP
🧠 NEXUS Experimental AI research environment exploring model-driven workflows and intelligent tooling. Python · AI · Research ↗ VIEW PROJECT	🧬 ML PIPELINE AGENT RL Experimental reinforcement-learning approach to ML pipeline decision-making. Python · RL · Q-Learning ↗ VIEW PROJECT
🛒 TRENDIFY Full-stack e-commerce product experience and web engineering build. React · Node.js · Express.js ↗ VIEW PROJECT	🛡️ MECHGUARD Mechanistic AI safety research across training and deployment. PyTorch · Interpretability · AI Safety ↗ VIEW PROJECT

ENGINEERING DNA

FULL STACK →

PROGRAMMING Python · C · C++ · JavaScript · TypeScript · SQL

ML / AI PyTorch · scikit-learn · XGBoost · Hugging Face Transformers · Deep Learning · Reinforcement Learning · Q-Learning · LSTM

GENERATIVE AI / AGENTS LLMs · Agentic AI · RAG · LangGraph · LangChain · Multi-Agent Systems · Prompt Engineering

DATA Pandas · NumPy · Matplotlib · Feature Engineering · Data Visualization

BACKEND / DEVOPS FastAPI · REST APIs · Docker · Git · GitHub · GitLab CI/CD · Linux

DATABASES / CLOUD SQLite · MySQL · PostgreSQL · Google Cloud · Microsoft Azure

OTHER React · Next.js · Node.js · Express.js · Streamlit · MQTT · OpenEnv

RESEARCH DIRECTION

MECHANISTIC AI SAFETY

Most AI interfaces expose:

INPUT ───────────────────────────────→ OUTPUT

My research asks what becomes possible when the internal state becomes observable.

REPRESENTATIONS Internal features and learned structure

ACTIVATIONS Signals inside the computation

GEOMETRY Changes across training and behavior

Core questions

Can internal geometric changes reveal important behavioral transitions?

Can activation-level signals expose coordination patterns in multi-agent systems?

Can safety monitoring move closer to mechanism rather than relying only on behavior?

Mechanistic Interpretability · Representation Geometry · Activation Analysis · Emergent Misalignment · Multi-Agent Safety · AI Monitoring

EXPERIENCE

2026 → PRESENT	EDGE DAEMON Research / venture direction · intelligent systems · mechanistic monitoring · AI safety
2026	PESU INNOVATION LAB Engineering + research within the university innovation ecosystem
2026	SINGLE CORE LABS Software / AI engineering experience

BUILDER RECORD

HACKATHONS / EVENTS 🔥 Scalar Grande Finale — Top Finalist 🔥 FantomCode — Round 2 Finalist ⚡ GitLab AI Hackathon ⚡ Meta × PyTorch OpenEnv ⚡ IEEE YESIST12 ⚡ IEEE Hack-Night

OPEN SOURCE GSSoC SSoC Hugging Face COMMUNITY IEEE RAS PESUCC Robotics Club IEEE EC Campus Kannada Koota

CREDENTIALS

Microsoft Azure — Generative AI & Neural Networks McKinsey Forward IIT Patna — AI / ML HackerRank — Python

Currently learning: Mechanistic Interpretability · AI Safety · Agentic Systems · Deep Learning · ML Systems

GITHUB / BUILD SIGNAL

LET'S BUILD

If you're working on AI systems, agents, ML infrastructure, interpretability, or AI safety, I'm always interested in ambitious problems and serious collaboration.

LinkedIn   ·   Hugging Face   ·   LeetCode   ·   CodeChef

_{BUILD · EXPERIMENT · UNDERSTAND · SHIP}