Building production-grade AI systems, semantic search infrastructure, computer vision pipelines, scalable backend applications and intelligent developer tools.

I build systems that sit between software engineering, machine learning, and data-heavy product work.

I am also interested in research on domains like AI-ML, data or systems related to them.

I like shipping real stuff, not toy notebooks. Search systems, retrieval pipelines, ML experiments, dashboards, browser extensions, OMR tooling, finance projects, and LLM-based workflows are my lane.

• Full stack web apps
• Machine learning and applied AI
• Research on interpretability, representation engineering, activation steering
• Search and retrieval systems
• Data analysis and analytics pipelines
• Computer vision and document automation
• Finance and market-related projects
• Browser extensions and developer tooling

Indian Institute of Technology Kharagpur

B.S. (Hons.) in Chemistry
M.Tech Dual Degree in Financial Engineering & Double Major in Industrial Engineering

2024 - 2029

Apr 2026 - Jul 2026

• Built production semantic search infrastructure using MiniLM, BM25 and Reciprocal Rank Fusion.
• Developed scalable PDF generation pipelines and optimized production OMR systems.
• Migrated backend services from Python to Node.js while improving search performance.

Nov 2025 - Dec 2025

• Fine tuned Vision Transformer and YOLO based segmentation models.
• Worked on dental X-ray segmentation under severe class imbalance.
• Improved training through mixed precision, augmentation and optimized scheduling.

Frameworks & Libraries

PyTorch • TensorFlow • Scikit-learn • Transformers • OpenCV • Keras • Hugging Face • XGBoost • LightGBM

Models & Architectures

BERT • SBERT • MiniLM • Vision Transformers (ViT) • SegFormer • YOLOv11 • Attention U-Net • TransUNet

LangChain • LangGraph • Prompt Engineering • Agentic AI • Semantic Search • Hybrid Retrieval • Vector Search

BM25 • Reciprocal Rank Fusion (RRF) • Embeddings • Cross-Encoder Re-ranking • ONNX Runtime

Libraries

NumPy • Pandas • SciPy • Statsmodels • Matplotlib • Seaborn • Plotly • NetworkX

Techniques

Exploratory Data Analysis • Feature Engineering • Statistical Analysis • Hypothesis Testing

Regression • Classification • Clustering • Time Series Forecasting • Probability & Statistics

Image Segmentation • Object Detection • Image Processing

OCR • Classical Computer Vision • Feature Extraction

Responsive UI • Component Architecture

REST APIs • WebSockets • Socket.IO

Authentication • Async Programming • Caching

Qdrant • IndexedDB • Vector Databases

GitHub Actions • MLflow

Google Colab • Kaggle

ETL Pipelines • Streaming Pipelines

Concurrent Processing • Caching

Web Scraping • BeautifulSoup

Mechanistic Interpretability

Representation Engineering

Activation Steering

Large Language Models

Retrieval Systems

Computer Vision

AI Systems

Multimodal Learning

Data Structures & Algorithms

Object-Oriented Programming

Operating Systems

Computer Networks

Database Management Systems

Software Engineering

Probability

Statistical Inference

Jupyter Notebook • GitHub Desktop • AWS • Google Cloud • Azure

• Hybrid-Multi-Agent-CRAG
• Context-Cast

• Behaviour-Score-for-Credit-Card
• Kaggle-Titanic-ML-From-Disaster
• House-Prices-Kaggle
• MckinseyDataAnalysis
• Summer-analytics-
• Trillytics-2026-Submission-
• NFL_AnalyticsSubmission

• FinLLM
• blackboxnlpAA
• Implementation-of-Hyena-Architecture-for-Transformers
• scientific-method-mining

• KerbSight
• Skin-Lesion-Analysis-and-Classification

• FlowState
• Alpha-Forge
• Vessa-SaaS-Typeshit
• Semantic-Memory-Extension

Some repos are private as they're under work or intern related.

• Representation Engineering
• Mechanistic Interpretability
• Activation Steering
• Retrieval Systems
• Large Language Models
• Computer Vision
• ML Systems

I like projects that are challenging in the right way at the start, then get engineered into something sharp.

This account mixes:

• experimental ML
• applied data work
• product prototypes
• search and ranking
• browser tooling
• finance and analytics
• real implementation over fake polish

• Codeforces Specialist (1300+)
• Solved 400+ DSA problems
• National Finalist, McKinsey Consularium
• JEE Main Top 0.9%
• JEE Advanced Top 6.7%

• stronger AI-ML systems
• Research related to interpretability, representation engineering, activation steering
• cleaner full stack architecture
• shipping projects that look real and work better than they look

• Stanford University Machine Learning Specialization
• DeepLearning.AI Neural Networks and Deep Learning

• GitHub: @ArnavLifelessCoder
• Email: arnavgawde1028@gmail.com
• Alternate: arnavgawde2810@gmail.com

Open to collaborations in AI, Machine Learning, Data Engineering, Search Systems and Full Stack Development.