My work sits at the intersection of AI systems, multimodal retrieval, orchestration, and full-stack product engineering. I like ambitious builds that are technically rigorous and visually polished.
AI Systems Portfolio
Most of my work lives around agentic AI, multimodal reasoning, deep learning, and interfaces that feel deliberate rather than generic.
I care about architecture, but I also care about how the final product feels. That balance shows up across my projects, from terminal-native agent tools to visual orchestration systems and research-driven ML work.
I design orchestration-heavy systems with memory, retrieval, tool use, and structured execution at the core.
My work spans multimodal RAG, reinforcement learning, knowledge distillation, model evaluation, and research-driven experimentation.
I like shipping the full stack, from interface systems and APIs to databases, infra, deployment, and developer tooling.
At Qpiai, I work on multi-agent infrastructure, model evaluation systems, and multimodal AI workflows across platform and computer vision initiatives.
Each project opens into a detailed case study alongside the GitHub repository.
A terminal-first AI agent platform built for speed, control, and codebase-aware context.
I built Forge-OSH as a modular Rust environment for AI-assisted terminal workflows, combining a polished TUI, provider routing, permissioned tools, and semantic code graph analysis.
A multimodal RAG system with knowledge graphs, adaptive retrieval, and layered memory.
Albot ingests text, images, audio, and video into an ArangoDB-backed knowledge graph, then answers complex queries through a retrieval stack that blends semantic, graph, and lexical signals.
A data science operating system for no-code and low-code experimentation, cleaning, and deployment.
DS-Forge turns common ML workflows into a full-stack platform with spreadsheet-style cleaning, rich feature engineering, model training, and auto-generated inference APIs.
A visual orchestration platform for agentic workflows, RAG pipelines, and code-first custom nodes.
Agflow brings agent workflow design into a visual canvas using React Flow, Supabase-backed storage, pgvector retrieval, external API execution, and custom Python extensibility.
A Jensen-Shannon divergence based distillation framework focused on stability and stronger compression.
This project explores how Jensen-Shannon divergence can outperform KL divergence for LLM distillation through smoother optimization, stronger dark knowledge transfer, and better downstream metrics.
Dynamic asset allocation research using DDPG, PPO, and embedding-driven RL on Indian equities.
I benchmarked multiple reinforcement learning strategies for portfolio management under transaction costs and market friction, then compared them with passive baselines on long-horizon NSE data.
A Transformer-based Spanish-to-English translation system trained on Europarl sentence pairs.
This project implements a full NMT pipeline with transformer architecture, subword tokenization, and beam search decoding for practical translation experiments.
A full-stack multi-agent system for subagent coordination, approvals, guarded execution, and tool-integrated workflows.
Alan Agentic Orchestrator is a multi-agent platform with a React and Next.js frontend, a FastAPI backend, multi-provider LLM routing, PostgreSQL and Redis-backed state, and external tool integrations through Composio.
An agentic research assistant for querying, comparing, and understanding CVPR papers with open-source models.
This project builds a local research assistant for CVPR literature using arXiv ingestion, FAISS retrieval, QLoRA-fine-tuned Qwen models, and a LangChain agent loop.
A Django-powered Akinator-style game driven by a decision tree classifier and a live feedback loop.
PrediNator is a guessing game with a decoupled machine learning core and Django interface. It asks yes, no, and don't know questions, makes interpretable predictions, and learns new characters through user feedback.
Grouped the way I actually use them in projects rather than as a flat logo wall.
Core languages I reach for across systems, AI, and web work.
Research and product tooling used in retrieval, modeling, and evaluation pipelines.
Interfaces, APIs, data layers, and product infrastructure.
Shipping, deployment, data processing, and dev workflows.
Computer engineering at NMIMS, supported by leadership work and community teaching.
Technical communities matter to me, not just the code itself.
Structured learning across AI, cloud, analytics, and engineering.
Orchestration, agent tooling, and systems that stay extensible as they grow.
Terminal-native agent tooling with multi-provider orchestration, secure execution, and code graph awareness.
A visual workflow builder for agentic AI with node-based composition, RAG, and code-first extensibility.
A full-stack multi-agent platform focused on decomposition, approvals, guarded execution, and tool-integrated workflows.
The fastest route is email, but all the main public links are here too.
