CV

Download my full academic CV here: Download CV

Education

• Bachelor of Technology in Computer Science and Engineering (AI & Data Science)

  • Sikkim Manipal Institute of Technology (SMIT), Sikkim, India
  • July 2023 – June 2027

• ISC (Class 11 & 12) – Science

  • St. Xavier’s School, West Bengal, India
  • April 2021 – March 2023

• ICSE (Class 1 – 10)

  • St. Xavier’s School, West Bengal, India
  • April 2010 – March 2021

Experience

• Founder & Chief AI Researcher

  Jan 2026 – Present

  • Halo Mind Research Group (Independent Research Organization)
  • Architectural Direction: Direct the fundamental scientific trajectory of an independent research group focused on hardware-efficient networks & sub-quadratic mechanisms.
  • Research Leadership: Manage a distributed team of ML Systems Engineers, Applied Vision Researchers, and Data Engineering Interns to execute end-to-end experimental designs.
  • Open-Source Development: Oversee the deployment and rigorous evaluation of robust PyTorch frameworks for space AI, solar-weather forecasting, medical imaging, and conformal prediction.

• Research Assistant (Satellite AI & Causal Inference)

  Jun 2026 – Present

  • AI & Global Development Lab (AIDevLab) – UT Austin / Chalmers
  • Spatial-Temporal Conformal Frameworks: Collaborating under the supervision of Prof. Connor T. Jerzak and Prof. Adel Daoud to engineer spatial-temporal conformal inference frameworks for satellite-imagery based poverty prediction models, demonstrating how standard conformal intervals break down under large geographic shifts.
  • Model Evaluations: Engineered custom Data-Loaders for continuous International Wealth Index (IWI) score prediction, establishing baseline prediction errors and conformal coverage metrics across geographic hold-outs (e.g., training on African data, evaluating on out-of-continent hold-outs).

• Research Intern (Computer Vision, OCR & Document Analytics)

  Jun 2026 – Present

  • Indian Statistical Institute, Kolkata
  • Geometric Recovery Architecture: Spearhead research under Prof. Umapada Pal to replace Thin-Plate Splines (TPS) with a sub-quadratic Vision Mamba (SSM) backbone, predicting dense 2D displacement fields to unwarp discontinuous metallic folds.
  • Forensic-Safe Semantic Recognition: Engineered a 3-stage pipeline integrating a Vision-Language Model (Qwen-VL) with a novel Conformal Risk-Controlled Abstention module, calculating normalized confidence scores to eliminate LLM hallucinations on destroyed texts.

• Research Intern (Computer Vision, OCR & Document Analytics)

  Jun 2026 – Present

  • The University of Salford, UK
  • Synthetic Dataset Engineering: Developed a scalable, automated dataset generation pipeline via Blender’s Python API (bpy) under Prof. Shivakumara Palaiahnakote, simulating extreme physical metal buckling and volumetric mud occlusion.
  • Mathematical Ground-Truth Extraction: Programmed the rendering pipeline to export uncompressed 32-bit OpenEXR multi-layer files, preserving pure floating-point vector coordinates for strict geometric regression.
  • Domain Generalization & Benchmarking: Curating a real-world physical “Crash-ALPR” test set to benchmark zero-shot domain adaptation against SOTA deformable rectifiers (e.g., ABINet, MORN), mathematically isolating geometric failure modes.

• Research Assistant & Team Lead

  July 2025 – July 2026

  • Sikkim Manipal Institute of Technology (SMIT)
  • Conducting research under Prof. Palash Ghosal in deep learning, focusing on Computer Vision, Bio-Medical, OCR & Reliable AI.
  • Medical Vision & Attention Mechanisms: Architected attention-enhanced Swin Transformer pipelines for complex medical diagnostics. Delivered robust brain tumor classification capabilities and generalized feature extraction under rigorous patient-level data splitting protocols (Accepted, IEEE GCON).
  • Fuzzy Logic & Medical Image Segmentation: Engineered AHF-RBF Net by replacing traditional Gaussian fuzzy membership with learnable Radial Basis Function (RBF) kernels. This architecture provides numerically stable, boundary-aware spatial attention for complex lesion segmentation, achieving a superior 84.55% IoU on the ISIC 2016 benchmark.
  • Cyber-Physical Security & Anomaly Rejection: Engineered Intrinsic Neural Firewalls utilizing Deep Delta Residual Overwrites for edge-deployed cyber-physical systems. Achieved high-performance, zero-shot anomaly rejection against False Data Injection Attacks (Accepted for Oral Presentation, WIN 6.0).
  • Research Leadership & Pipeline Engineering: Directed and mentored student research teams, training junior researchers in core deep learning methodologies and guiding end-to-end experimental design from conceptualization to multiple first-author and co-authored acceptances in IEEE and Springer venues.

• AI & Data Science Intern

  Jul 2025 – Aug 2025

  • Soft Nexis Technology
  • Predictive Analytics Pipeline: Engineered and deployed RealVisor, an end-to-end AI real estate platform. Developed robust predictive pipelines utilizing XGBoost and Random Forest Regressors to estimate property valuations based on complex spatial features.
  • Model Evaluation & Deployment: Conducted rigorous evaluation across curated real-world datasets, strictly outperforming baseline models, and designed a production-ready Streamlit dashboard featuring interactive market trend visualizations and automated investment analysis.

Honors, Certifications & Academic Service

• Technical Peer Reviewer: Invited and served as an official peer reviewer for the IEEE GCON conference, evaluating manuscripts in applied deep learning and computer vision.

• Top 1% Topper & Gold Medalist: Introduction to Internet of Things, NPTEL (Funded by MoE, Govt. of India). Elite score: 91%.

• Deep Learning & AI Certifications: NLP with DL, PyTorch for Medical Imaging (Udemy); GenAI for Data Scientists (Coursera); Applied Machine Learning (AMII).

Technical Skills

• Languages & Frameworks: Python, C, Java, PyTorch, TensorFlow, Keras, Blender, Streamlit, LaTeX, Git, GitHub

• Computer Vision: Vision Transformers, Mamba, State Space Models (SSMs), U-Net, Deep Delta Learning, Diffusion Models, GANs, YOLO, CNNs

• Signal Processing & Time Series: 1D CNNs, Signal Analysis, SpaCy, NLTK, Text Mining, OCR Pipelines

• Data Engineering & Tools: Pandas, NumPy, SciPy, Scikit-learn, ETL Workflows, Hadoop, Spark, Computational Workflows, High-Throughput Image Processing