Scroll
AI Researcher & Engineer

Abhinav Shukla

B.Tech CSE (AI) @ CSVTU Bhilai · Research Intern @ ISI Kolkata CVPR + IIIT-H MLL · NeurIPS 2026 + ICDAR 2026 submissions
Building at the intersection of 3D Computer Vision, Explainable AI, and Machine Unlearning.

View Projects Get In Touch
200K+
PyPI Downloads
3
Active Research Labs
20+
Workshops Organized
4
Papers Submitted
01 // Experience

Where I've
Worked & Researched

JAN 2026 — Present
Research
Machine Learning Lab, IIIT Hyderabad
Research Intern · 3D Computer Vision · Dr. Charu Sharma
Conducting research in 3D Computer Vision at one of India's premier AI research labs under Dr. Charu Sharma.
DEC 2025 — Present
Research
University of Salford, UK
Research Intern · Dr. Shivakumara P.
International research collaboration on computer vision and document analysis problems.
OCT 2025 — Present
Research
Indian Statistical Institute, Kolkata
AI/ML Research Intern · CVPR Unit
Conducting research under Dr. Umapada Pal (IAPR Fellow) and Dr. Shivakumara P. (Univ. of Salford, UK). Resulted in an ICDAR 2026 submission on scene text spotting.
APR — JUL 2025
Industry
NIT Jamshedpur / IIIT Naya Raipur
AI/ML Research Intern
Designed a deep feature-based particle filter tracking system combining YOLOv8 + MobileNetV2 with multi-objective optimization (NSGA-II, MOPSO) for real-time people tracking in dynamic environments.
JAN 2026 — Present
Academic
Springer Nature
Reviewer
Peer reviewing research submissions for Springer Nature publications.
MAR — JUL 2025
Open Source
Social Summer of Code 2025
Project Admin & Mentor
Mentored contributors across multiple AI/ML open-source projects. Conducted workshops on LLM fine-tuning, AI frameworks, and open-source contribution practices.
02 // Projects

Things I've
Built

PROJECT_001
200K+ Downloads
Annie — Rust-Powered ANN Library
Lightning-fast Approximate Nearest Neighbor search library for Python. SIMD-accelerated brute-force, HNSW indexing, GPU acceleration, zero-copy NumPy integration. 200K+ PyPI downloads.
RustSIMDHNSWGPUPyPI
PROJECT_002
Claude Computer Use Agent
Containerized backend for AI desktop automation. FastAPI + SQLAlchemy + Docker + WebSockets for real-time orchestration. VNC-powered virtual desktop with live monitoring.
FastAPIDockerWebSocketsAnthropic
PROJECT_003
End-to-End MLOps Streaming Pipeline
Real-time image classification with PyTorch ResNet served over gRPC. Kafka for streaming video frames, Docker containerization, Streamlit dashboard for one-click system control.
KafkagRPCPyTorchMLOps
PROJECT_004
Explainable GeoAI Pipeline
Spatial ML pipeline predicting U.S. county vote-share (R²≈0.90) using FLAML-tuned XGBoost. Tree SHAP + GeoShapley decomposition with interactive Folium + Plotly dashboard.
XGBoostSHAPGeoPandasFLAML
PROJECT_005
Full World Model (Vision + Memory + Controller)
World Model with CNN encoder, LSTM memory, and controller for autonomous driving in CarRacing-v3. Latent state prediction, RL rewards, automated evaluation with video logging.
Model-Based RLLSTMCNNPyTorch
PROJECT_006
XFeat — CVPR 2024 Implementation
From-scratch implementation of XFeat (CVPR 2024) optimized for efficiency and scalability in real-world image matching. Benchmarked across datasets with deployment-ready optimization.
Computer VisionPyTorchCVPR
PROJECT_007
FATA — WACV 2025 Implementation
Implementation of Feature Augmentation based Test-Time Adaptation from WACV 2025. Robust domain adaptation under distribution shifts with interactive Streamlit demonstration.
Domain AdaptationTTTWACV
03 // Research

Papers &
Publications

NeurIPS 2026 · #9372
Taylor-Made Sparsity: First-Order Approximation for Structural Pruning in Object Detectors
Structural pruning framework leveraging first-order Taylor approximations for efficient object detection. Co-authored with Shefali Bajaj & Nachiket Tapas.
Under Review
NeurIPS 2026 · #11046
Selective Unlearning in Multi-Agent Deep Deterministic Policy Gradient
Machine unlearning in multi-agent reinforcement learning via selective forgetting in MADDPG. Co-authored with Anoushka Chatterjee, Faiz Ali & Nachiket Tapas.
Under Review
ICDAR 2026
Reliability-Weighted Semi-Supervised End-to-End Approach for Scene Text Spotting
Developed during research at ISI Kolkata CVPR Unit under Dr. Umapada Pal and Dr. Shivakumara P. Advances robust scene text spotting under semi-supervised settings.
Under Review
ICDAR 2026 · arXiv:2602.14040
Explainability-Inspired Layer-Wise Pruning of Deep Neural Networks for Efficient Object Detection
Novel SHAP-based pruning framework achieving superior accuracy–efficiency trade-offs for object detection in resource-constrained environments.
Under Review
NITRR Seminar 2024
EcoLens: A Visionary Solution for E-Waste Management through Intelligent Imaging
Computer vision–based approach for e-waste detection and classification achieving 85% accuracy. Promotes sustainable practices aligned with UN SDG 12.
Presented
04 // Academic Services

Peer Review &
Community

JAN 2026 — Present
Reviewer
Springer Nature Computer Science
Peer Reviewer · Impact Factor: 2.185
Peer-reviewing manuscripts in Artificial Intelligence, Computer Vision, and Machine Learning.
JAN 2026 — Present
Reviewer
Artificial Intelligence & Applications (AIA)
Peer Reviewer · BonView Press · Impact Factor: 9.375
Peer-reviewing manuscripts in core AI and applications domains.
05 // Capabilities

Technical
Arsenal

// Languages
PythonRustC/C++ JuliaRBash
// ML / Deep Learning
PyTorchTensorFlowKeras OpenCVScikit-LearnGymnasium CNNsLSTMsDDPG
// LLMs & Gen AI
HuggingFaceLangChainPEFT/LoRA UnslothChromaDBSFTTrainer
// DevOps & Cloud
DockerFastAPIGitHub Actions CI/CDPrometheusGrafana KafkagRPC
// Data Science
NumPyPandasGeoPandas SHAPFLAMLGeoShapley PlotlyStreamlit
// Databases
PostgreSQLSQLite FirebaseChromaDB
06 // About

The Human
Behind The Code

Started programming in 3rd grade. Now conducting research at ISI Kolkata, IIIT-H MLL, and University of Salford simultaneously — as a 2nd year undergrad.

President of Chhattisgarh's most active coding club. Creator of Annie (200K+ downloads). Reviewer for Springer Nature and AIA (IF: 9.375). Rank 19 at International Informatics Olympiad 2019.

NASA Citizen Scientist. HackerRank 5-Star. Always building, always publishing.

DeepLearning.aiLinear Algebra for ML & DS
Duke UniversityExplainable AI (XAI)
Duke UniversityRust for DevOps
Univ. of AlbertaReinforcement Learning
IBMDeep Learning with PyTorch
IBMMachine Learning with Python
abhinav@research:~$
❯ whoami
abhinav_shukla // AI researcher + engineer

❯ cat status.json
{
  "degree": "B.Tech CSE(AI) @ CSVTU",
  "year": 2,
  "labs": [
    "ISI Kolkata",
    "IIIT-H MLL",
    "Univ. of Salford"
  ],
  "papers": 4,
  "downloads": "200K+"
}

❯ ls interests/
computer_vision/  explainableAI/
high_perf_systems/  rl/

❯ echo $AMBITION
advance AI research. build things that matter.

❯
07 // Contact

Let's Build
Something Extraordinary

Open to research collaborations, internships, and interesting problems.

Email Me LinkedIn GitHub +91 7724033472