DEVELOPER_PROFILE

Applied ML with a Systems Engineering Mindset

B.Tech Computer Science student specializing in AI & ML at Siliguri Institute of Technology (MAKAUT), expected graduation 2028. Currently interning as an AI/ML Developer at Hansraj Ventures — building production RAG pipelines, LoRA fine-tuning workflows, RAGAS evaluation harnesses, and high-concurrency async microservices.

TECH_STACK

Categorized Engineering Stack

From retrieval and adapter fine-tuning to real-time observability and production async APIs.

AI/ML & LLM Systems

RAG Pipelines Hybrid Search (BM25 + Milvus/FAISS/Qdrant) LoRA Fine-Tuning Adapter Merging Hallucination Guardrails RAGAS Evals Langfuse Tracing Ollama · Local LLMs HuggingFace PEFT

Applied Machine Learning

Regression Modeling Time-Series Forecasting Anomaly Detection CNNs Transfer Learning Computer Vision Pandas NumPy Scikit-Learn

Languages & Frameworks

Python FastAPI PyTorch TensorFlow Flask LangChain Java SQL JavaScript HTML/CSS

Systems, Infra & Core CS

Async Microservices Docker CI/CD GCP MongoDB PostgreSQL Vector DBs (Milvus · FAISS · Qdrant) Git Data Structures & Algorithms System Design

MILESTONES

Track Record & Accomplishments

WINNER

Smart India Hackathon (Internal Round) — Civic Connect CNN

1st RUNNER-UP

45-Day DSA Challenge — Top algorithmic performance

3rd PLACE

SIT Code Quest — Competitive programming

200+

LeetCode Problems Solved — 100 Days Badge earned

ACADEMIC_FOUNDATION

Education & Certifications

Siliguri Institute of Technology (MAKAUT)

B.Tech in Computer Science & Engineering — AI & ML Specialization

Expected: July 2028
CGPA: 7.2

Professional Certifications

Mastering Generative AI & ChatGPT (Scaler) · AWS APAC Solutions Architecture Simulation (Forage) · Accenture Nordics Software Engineering Simulation (Forage)

Verified Credentials

Open Source Contributions

GirlScript Summer of Code (GSSoC) 2024 — contributed bug fixes, performance improvements, and documentation to developer tooling and ML repositories.

Jul–Nov 2024

CORE_PHILOSOPHY

Engineering Principles

How I approach building AI systems that survive in production rather than breaking on edge cases.

01. Evals Before Scaling

MEASUREMENT

Every LLM or fine-tuned adapter is coupled with automated RAGAS test harnesses measuring faithfulness, relevancy, and context recall before deployment. What isn't measured drifts.

02. Zero Hallucination Tolerance

SAFETY

In legal and financial domains, fabricated outputs cause real liability. I build deterministic citation cross-checkers and strict cosine guardrails into the pipeline flow.

03. Low-Latency Async I/O

SYSTEMS

Async FastAPI microservices, non-blocking DB pools, Milvus hybrid search with reciprocal rank fusion, and quantized CUDA inference ensure sub-second p95 latencies.