I specialize in architecting Multi-Agent systems, stateful LLM workflows, and scalable deep learning pipelines. As a Dual Degree (B.Tech + M.Tech) graduate in Ocean Engineering and Naval Architecture from IIT Kharagpur (CGPA: 8.12), my focus is on bridging theoretical algorithmic problem-solving with deterministic, production-grade AI systems.
- 🔭 I’m currently focused on: Multi-Agent Orchestration (LangGraph), Long-Term Memory Systems, and Agentic RAG.
- 🌱 I have expertise in: Deep Learning (TensorFlow/Keras), LangChain, and orchestrating complex ML pipelines.
- 🏆 Milestones: Secured All India Rank (AIR) 64 in GATE 2024 | Recipient of the Prof. J.P. Ghose Memorial Award | Gold Medalist in Gymkhana Championship Choreography.
- 👨🏫 Leadership: Guided 500+ aspiring engineers at AlgoZenith in Data Structures, Algorithms, and interview problem-solving.
- 📫 How to reach me: sumanraj4176@gmail.com | Portfolio | LinkedIn
-
Generative AI & LLMs:
-
Machine Learning & Deep Learning:
-
Programming Languages:
-
Frameworks and Libraries:
-
Platforms & Tools:
-
Core Engineering & Fundamentals:
-FF4F8B?style=flat-square){kind=icon}
-2E8B57?style=flat-square){kind=icon}
- Tech Stack:
PythonLangGraphLangChainOpenAITavilyWikipediaLangSmith - Built a multi-agent research system using LangGraph to decompose complex topics into specialized analyst perspectives with human-in-the-loop refinement.
- Implemented parallel information retrieval using Tavily and Wikipedia, enabling agents to gather and synthesize context from multiple sources.
- Implemented a map-reduce pipeline to parallelize analyst interviews and synthesize a comprehensive final report.
- Tech Stack:
PythonLangGraphLangChainOpenAIPydanticTrustcall - Built a stateful AI task-management agent using LangGraph that maintains short-term conversational state and structured long-term memory across interactions.
- Designed separate Profile, To-Do, and Instruction memory with Pydantic schemas, enabling structured memory extraction, updates, and personalized task management.
- Implemented conditional agent routing and cyclic workflows to determine when memory updates are required, route operations to the appropriate memory node, and return control to the main agent.
- Tech Stack:
PythonTensorFlowKerasMobileNetV3-SmallOpenCV - Built a Presentation Attack Detection (PAD) system using MobileNetV3-Small transfer learning to classify fingerprint images as LIVE or SPOOF for biometric authentication.
- Developed an end-to-end pipeline covering dataset splitting, image augmentation, ImageNet normalization, model training, threshold calibration, and single-image inference.
- Calibrated the threshold against a target BPCER of 3% and evaluated held-out test performance using APCER, BPCER, ACER, EER, ROC-AUC, F1-score, and APCER-BPCER trade-off analysis.
- Tech Stack:
PythonPandasNumPyScikit-learnTensorFlow/Keras - Built an end-to-end credit risk prediction pipeline using TensorFlow/Keras on 396K+ Lending Club loan records, covering missing-value handling, categorical encoding, feature engineering, and MinMax scaling.
- Trained a deep neural network with ReLU activations, dropout, and sigmoid output to classify loans as fully paid or defaulted.
- Achieved a 0.93 F1-score on the test set, effectively handling significant class imbalance in loan default prediction.
- Created a Python profiling package on PyPI for function tracing and execution-time analysis to identify performance bottlenecks.
-
AI/ML Engineer Intern | MakeMyBrain
(May 2024 – Jul 2024)- Built an end-to-end mood-based music recommendation RAG system, processing structured metadata, acoustic features, and textual profiles across audio catalogs.
- Engineered a hybrid search pipeline combining BM25 keyword matching and dense vector embeddings (
$\alpha=0.4$ ) boosting retrieval Recall@10 from 0.61 to 0.79 compared to lexical search alone. - Implemented a cross-encoder re-ranking stage to filter top 100 retrieved candidates down to top 10-20 tracks before LLM generation, ensuring personalized and context-aware recommendations.
-
Teaching Assistant | Indian Institute of Technology, Kharagpur
(Aug 2023 – May 2025)- Conducted tutorials, evaluated assignments, and led doubt-clearing sessions for undergraduate courses at IIT Kharagpur.
-
Mentor | AlgoZenith
(Oct 2022 – May 2024)- Mentored 500+ students in Data Structures and Algorithms, guiding them through complex coding challenges and technical problem-solving.