- 🎓 Data Science and Applications student at IIT Madras, with a strong focus on software engineering, AI/ML, and full-stack product development.
- 🤖 Build LLM-orchestrated systems with multi-provider routing, retrieval-augmented generation (RAG), and grounding/validation pipelines.
- 🧩 Deliver full-stack applications end-to-end using Flask, FastAPI, React, Vue.js, REST APIs, SQLAlchemy, and relational databases.
- 📊 Apply a product and analytics mindset through feature engineering, forecasting, data analysis, and BI-style reporting.
- ⚙️ Comfortable across database design, backend architecture, workflow automation, and Git-based collaboration.
- 🌱 Continuously exploring generative AI, cloud computing, and scalable data systems through coursework and projects.
🎯 Open To: Software Engineering · AI/ML Engineering · Full-Stack Development · Data Science Roles & Internships
Languages
Frontend
Backend & Databases
Cloud, DevOps & Tooling
| Domain | Proficiency | Details |
|---|---|---|
| LLM Orchestration | ⭐⭐⭐⭐☆ | Multi-provider routing with role-based model selection, retry/failover handling, and structured execution flows |
| Retrieval-Augmented Generation (RAG) | ⭐⭐⭐⭐☆ | Source-grounded pipelines with claim matching and validation to reduce hallucinations |
| Machine Learning | ⭐⭐⭐⭐☆ | Scikit-learn, XGBoost, LightGBM, forecasting, regression, and feature engineering |
| Prompt Engineering | ⭐⭐⭐⭐☆ | Role-based prompts, structured outputs, and schema-constrained generation |
| Multi-Agent Systems | ⭐⭐⭐☆☆ | Coordinator-style patterns for extractor, planner, generator, and auditor roles |
| Data Analysis & BI | ⭐⭐⭐⭐☆ | Excel financial modeling, pivot tables, Power BI dashboards, and KPI analysis |
🧩 Gyantra — AI Orchestration Pipeline for Education
Transforms educational documents into structured, source-grounded Teacher Knowledge Packages, including lesson plans, classroom activities, assessments, and misconception analysis.
| Aspect | Detail |
|---|---|
| Stack | FastAPI, React, Vite, SQLite, Docker, Gemini / Groq / OpenRouter |
| Scale | 10-stage async orchestration pipeline with per-role model routing |
| Performance | SSE-based live progress streaming and disk-cached LLM responses |
| Security | Origin-tagged content with two-pass grounding and hallucination checks |
| Impact | Converts a chapter upload into a teacher-ready package in minutes |
| Repository | github.com/Hariomkr147/Gyantra |
Built as a fixed, auditable pipeline instead of a general-purpose agent framework so each stage can fail gracefully while critical checks still block bad outputs.
🎬 Cinema Audience Forecasting
Machine learning project to forecast cinema audience demand from historical sales and attendance data.
| Aspect | Detail |
|---|---|
| Stack | Python, Pandas, NumPy, Scikit-learn, Matplotlib |
| Scale | Multi-feature historical dataset with engineered demand indicators |
| Performance | Benchmarked multiple regression and classification models for forecast accuracy |
| Security | Input validation and preprocessing checks for data consistency |
| Impact | Produced actionable forecasts for scheduling and inventory decisions |
| Repository | github.com/Hariomkr147/Cinema-Audience-Forecasting-challenge |
🎓 Placement Portal — Full-Stack Recruitment Platform
Role-based placement management platform for Admin, Company, and Student workflows.
| Aspect | Detail |
|---|---|
| Stack | Flask, Vue.js, SQLite, REST APIs |
| Scale | Three-role platform with company onboarding and job-posting approval flows |
| Performance | RESTful API-driven frontend/backend integration |
| Security | Role-based access control across all workflows |
| Impact | Streamlined campus recruitment operations |
| Repository | github.com/Hariomkr147/placement-portal |
💧 BDM Capstone — Business Analytics Case Study
Business analytics case study built around cleaning raw sales data and turning it into strategic recommendations.
| Aspect | Detail |
|---|---|
| Stack | Excel (Pivot Tables, Financial Modeling), Power BI |
| Scale | 61 days of unstructured sales/pricing data across 5 SKUs |
| Performance | Automated revenue, profitability, and fixed-cost calculations |
| Security | N/A — offline business dataset |
| Impact | Profitability benchmarking informed pricing and diversification recommendations |
| Repository | github.com/Hariomkr147/bdm-project |
IITM Software Engineering · Current
Civic complaint-management platform built with a five-member Scrum team.
- Designed normalized database tables for users, zones, complaints, evidence, points, badges, and complaint history
- Implemented geo-proximity complaint grouping using the Haversine formula with zone-based routing
- Built complaint lifecycle workflows, audit trails, and role-based access control
- Collaborated via GitHub issues, pull requests, and code review in a Scrum workflow
Flask SQLAlchemy REST APIs SQLite Scrum GitHub
| Recognition | Details |
|---|---|
| 🥇 Winner — IPL Auction Showdown 4.0 | IITM Paradox 2026 |
| 🥈 Runner-up — IPL Arena | IITM Paradox 2024 |
| 🎖️ Participant — DataVista Analytics Competition | IITM Paradox 2025 |
| 🎖️ Participant — TradeQuest 2024 | IIT Madras |
IIT Madras
HackerRank
Independent Courses
Requires the
Platane/snkGitHub Action on this profile repo to generate theoutputbranch SVGs.
Learning: Advanced LLM Orchestration, Multi-Agent Systems, Distributed Systems
Building: AI-powered education tools & business analytics platforms
Exploring: Retrieval-Augmented Generation, Cloud-Native Architectures
Open To: Software Engineering Roles · AI/ML Engineering Roles · Full-Stack Opportunities
"Build things that make someone else's work simpler, safer, or faster."