Anindya Majumder

AI EngineerLecturerResearcher

I build retrieval systems that companies can trust, teach algorithms at BRAC University, and test what vision models learn from bad labels.

Résumé ↗Email GitHub LinkedIn Medium

Experience

BRAC University
Adjunct LecturerAug, 2026 - Present
- Designing and delivering lectures on Data Structures and Algorithms to undergraduate Computer Science students.
- Setting and evaluating exams, assignments, and quizzes, and mentoring students through consultation hours on algorithmic problem-solving and complexity analysis.
Data Structures · Algorithms · Complexity Analysis · Java · Python
Undergraduate Teaching AssistantJune, 2024 - May, 2025
- Tutored Data Structure and Object Oriented Programming to over 100+ students.
- Provide consultation, marked quizzes and assignments, and assisted in lab sessions.
Java · Object Oriented Programming · Data Structures · Algorithms
RiseUp Labs Ltd.
AI Workflow Automation EngineerApril, 2026 - Present
- Engineering LLM-powered automation pipelines for marketing, sales, lead generation, and project management teams, reducing manual operational overhead.
- Architecting role-based multi-agent RAG system over the company's vast central database, enabling access-controlled, context-aware query responses tailored to each user's role.
- Deployed and monitoring hybrid AI inference architecture with locally served open-source models via vLLM for sensitive workloads and cloud APIs for general tasks, optimizing cost, latency, and data security.
- Conducting research on RLHF, DPO, and GRPO techniques to improve behavioural alignment, reasoning, safety, and performance optimization of generative AI models.
LangChain · LangGraph · Claude Code · Git & Github · VectorDB · Docker
Join Venture AI
Junior AI DeveloperOct, 2025 - Feb, 2026
- Developed and integrated Natural Language Query (NLQ) features into systems, allowing AI agents to interact with data from PostgreSQL.
- Implemented Containerization and CI/CD pipelines using GitHub Actions to automate testing, building, and deployment processes.
- Optimised LLM token usage by prompt caching, semantic re-ranking, and dynamic model routing, significantly reducing production costs.
- Fine-tuned and distilled large generative models to create efficient, niche-specific variants with reduced latency and lower resource consumption.
- Engineered a multi-agent system for an English vocabulary learning platform, featuring real-time user progress tracking and agentic database operations.
- Developed custom MCP servers to securely connect AI agents with internal databases and external APIs for seamless data retrieval.
Python · FastAPI · Git & Github · LangGraph · LangChain · VectorDB · PostgreSQL · Docker
Trainee AI DeveloperJul, 2025 - Sep, 2025
- Implemented agentic graph-RAG solutions with a re-ranker for legal chatbot applications.
- Contributed to conversational AI systems focused on mental health support.
- Multi-agent RAG-powered workout plan generation based on user preference and the dietary chart.
- Developed multi-agent RAG systems for order and reservation through phone call, for SAAS restaurant receptionist.
Python · FastAPI · PostgreSQL · LangChain · VectorDB · LangGraph · Agent Development Kit(ADK)

Research

Sentinel-2 shortwave-infrared view of a wildfire burn scar, shown dark against surrounding vegetation

Do Geospatial Foundation Models Learn Robustly from Imperfect Labels?

Five pretrained encoders, HLS Burn Scars and FLAIR-2, four kinds of label error. Two multispectral foundation models showed no robustness advantage over an ImageNet ResNet-50.

Under Review · WACV 2027 GeoCV Workshop

Laparoscopic frame with the model's segmentation masks over the stomach, colon and abdominal wall

Hierarchical Transformer-Based Semantic Segmentation of Intraoperative Anatomical Structures

A DINOv2 backbone with a multi-scale cross-attention decoder lifted Dice by up to 27% on hard organ classes like the pancreas and colon.

In Progress · BRAC University ↗

Projects

Architected a GraphRAG pipeline with LangChain, Neo4j, Cypher, FastAPI, and LangGraph, combining entity linking, subgraph expansion, hybrid retrieval, and reranking for citation-grounded response generation over a legal knowledge graph. Established evaluation and observability with RAGAS, tracking faithfulness, relevance, F1 score, and retrieval quality to harden response reliability on ambiguous queries. Optimised local inference using GGUF/4-bit quantisation, KV cache, FlashAttention, and speculative decoding, achieving faster generation with reduced memory footprint.
AI Lawyer
A GraphRAG pipeline for citation-grounded response generation over a legal knowledge graph, featuring local inference optimization.
Python · LangChain · Neo4j
Built AI voice agents with intent routing, tool-calling, and stateful dialogue for fully automated reservations, ordering, and customer support over phone. Integrated Twilio and Vapi for speech-to-speech pipelines, implementing event-driven call routing, spam filtering, and human escalation with async callback automation. Developed OCR-based menu ingestion using vision-language models for structured extraction across multi-modal restaurant data pipelines.
Open ↗
TrustTaste
AI voice agents and automated systems for restaurant reservations, ordering, and customer support.
Python · Twilio · Vapi
Developed scalable FastAPI backend with microservice architecture, dependency injection, and layered design for maintainability and independent service scaling. Built AI-powered voice assistant using ElevenLabs Client SDK to guide users, teach vocabulary, and evaluate learning through AI-generated quizzes. Optimised performance through Redis caching, rate limiting, memoisation, and database query tuning, reducing response latency across high-traffic endpoints. Containerised with Docker and deployed via CI/CD pipelines on GitHub Actions to VPS.
Open ↗
Colwords
A scalable vocabulary platform featuring AI-powered voice assistants and microservice architecture.
FastAPI · Python · ElevenLabs
Built scalable backend with Spring Boot applying layered and domain-driven design for modularity and separation of concerns. Architected stateless JWT authentication with RBAC, enabling horizontally scalable session management and secure access control. Optimised MongoDB persistence layer with indexing, connection pooling, and pagination to support high-throughput ingestion and retrieval. Engineered real-time notifications via Server-Sent Events (SSE), decoupling read-heavy streams from core paths to improve overall responsiveness.
Source ↗
NAIS Journal
A scalable micro-blogging platform with real-time notifications and stateless JWT authentication.
Spring Boot · MongoDB · JWT