Hirakjyoti Basumatary
Research Area: Robotic Grasping, Machine Learning and Adaptive Control
Publications :
- Hirakjyoti Basumatary, and Shyamanta M. Hazarika."Design optimization of an underactuated tendon-driven anthropomorphic hand based on grasp quality measures." Robotica (2022): 1-20.
- Hirakjyoti Basumatary, and Shyamanta M. Hazarika." State of the Art in Bionic Hands." IEEE Transactions on Human-Machine Systems, 50, no. 2 (2020): 116-130.
- Dey, Abhijit, Basumatary, Hirakjyoti, and Shyamanta M.Hazarika. "A Decade of Haptic Feedback for Upper-Limb Prostheses ." IEEE Transactions on Medical Robotics and Bionics (2023)
- Hirakjyoti Basumatary, Daksh Adhar, Atharva Shrawge, Prathamesh Kanbaskar and Shyamanta Moni Hazarika "A Reinforcement Learning Approach for Bionic Reflex Control in Anthropomorphic Hands Exploiting Domain Randomization," Robotics and Autonomous Systems, Elsevier, 2024. (Revision Submitted)
- Hirakjyoti Basumatary, Tamen Thapa Sarkar and Shyamanta Moni Hazarika "Robustifying a Reinforcement
- Learning based Grasped Object Slippage Prevention Controller through Adaptive Sliding Mode Control," 28th International Conference on Methods and Models in Automation and Robotics (MMAR) 2024. (Submitted)
- Hirakjyoti Basumatary, Tamen Thapa Sarkar and Shyamanta Moni Hazarika "Robustifying a Reinforcement
- Learning Agent based Bionic Reflex Controller through an Adaptive Sliding Mode Control," Robotica 2024.
- (Submitted)
- Hirakjyoti Basumatary, Atharva Shrawge, Prathamesh Kanbaskar and Shyamanta Moni Hazarika "Low-Level Grasp Controller for Slippage and Deformation Prevention exploiting Deep Reinforcement Learning" In Advances in Robotics-6th International Conference of The Robotics Society, 2023
- Hirakjyoti Basumatary, Daksh Adhar, Riddhiman Shaw, and Shyamanta Moni Hazarika "Grasp Force Optimization as a Bilinear Matrix Inequality Problem: A Deep Learning Approach" 6th National Conference on Multidisciplinary Design, Analysis and Optimization (NCMDAO 2023)
Key courses taken:
- Machine Learning: Pattern Recognition and Machine Learning, Reinforcement Learning, Deep Learning, Meta Learning, Transfer Learning, Deep Unsupervised Learning, Recommender Systems, MLOps Bootcamp (Udemy)
- Natural Language Processing: Introduction to Natural Language Processing, Natural Language Processing in Python, Natural Language Processing with Transformers, CS224N NLP with Deep Learning
- Robotics: Robotics and Robot Applications, Mechanics and Control of Robot Manipulators, Robotics Operating System (ROS 1 and ROS 2), Mechatronics, Signal Processing, Wheeled Mobile Robotics
- Control: Control Systems, Nonlinear Adaptive Control, Sliding Mode Control, Digital Control
- Computer Vision: Deep Learning for Computer Vision, Computer Vision Basics, Modern Computer Vision
- Miscellaneous: Object Oriented Programming in Python, Search Methods in Artificial Intelligence, Fundamentals of Artificial Intelligence, Data Structures and Algorithms, Finite Element Analysis
- Role
Fellow, Principal ML & Robotics Engineer
- Years of Experience
10 years
Skillsets
- Python - 3 Years
- PyTorch - 5 Years
- TensorFlow - 5 Years
- C++
- Docker
- FastAPI
- Kubernetes
- ROS
- Arduino
- AutoCAD
- AutoGen
- Aws sagemaker
- Azure ML
- Beeai
- C
- crewAI
- Crocoddyl
- CSS
- Cura
- Cypher
- Flask
- Ftetwild
- Gazebo
- GitHub Actions
- Gmsh
- Grafana
- Graspit
- HTML
- Hugging Face Transformers
- Isaac lab
- Isaac ros
- Isaac Sim
- JavaScript
- JAX
- Keras
- Kubeflow
- LangChain
- LangGraph
- LangSmith
- LaTeX
- Linux
- LlamaIndex
- Meshcat
- Microsoft Office
- MLFlow
- MongoDB
- Mujoco
- MySQL
- Neo4j
- NLTK
- NoSQL
- NumPy
- NVIDIA Omniverse
- OpenCV
- pandas
- Pinnochio
- PostgreSQL
- Prometheus
- Pybullet
- Pydantic
- Raspberry Pi
- Rviz2
- Scikit-learn
- SciPy
- SolidWorks
- spaCy
- SPARQL
- SQL
- Stable-baselines3
- Streamlit
- Syngrasp
- Tableau
- Tetwild
- Vertex AI
- Viser
- Voxelizer
- Windows
- Yolo
- ChromaDB
- ComfyUI
- MATLAB
- MCP
- Milvus
- MLOps
- n8n
- Pinecone
- Qdrant
- Simulink
- Skrl
- TimescaleDB
- Amazon Web Services
- Apache Airflow
- Apache Flink
- Apache Hudi
- Asana
- Azure
- Azure DevOps
- Balsamiq
- Confluence
- Data Lakehouse
- Dovetail
- Druid
- Figma
- Google Analytics
- Google Cloud
- Hubspot
- Jira
- Kafka
- Miro
- Mixpanel
- Notion
- Nvidia Jetson
- PySpark
- Surveymonkey
- Trello
Professional Summary
10Years
- Sep, 2025 - Present 6 months
Fellow & Principal ML Engineer
Rubric Academy - Nov, 2025 - Dec, 2025 1 month
Neurosymbolic AI Architect
Aiql - Jan, 2025 - Oct, 2025 9 months
Founding Robotics Lead
Machines For Humans - Jan, 2017 - Jan, 20247 yr
PhD Research Scholar/Assistant
Indian Institute Of Technology, Guwahati - Jul, 2024 - Jan, 2025 6 months
Senior Robotics Engineer
Kion Group
Applications & Tools Known
Python
C++
C
Keras
PyTorch
Scikit-Learn
Pandas
NumPy
SciPy
MLOps
OpenCV
YOLO
LangChain
NLTK
LLMOps
Amazon Web Services
Azure
.png)
Docker
MLFlow
AutoCAD
Arduino
Raspberry Pi
Microsoft Office
LaTeX
Google Cloud Platform
Work History
10Years
Fellow & Principal ML Engineer
Rubric AcademySep, 2025 - Present 6 months
Selected as a Fellow to architect 'Golden Response' datasets and RLHF benchmarks for aligning Frontier Agents on long-horizon reasoning tasks (Project Synapse/MLE-Bench). Designed dependency-based reasoning graphs for complex query decomposition, engineered multi-hop reasoning benchmarks, defined 'Code-as-Policy' evaluation standards, structured hierarchical grading rubrics for reward models, developed adversarial system prompts and datasets, curated human-verified SFT datasets, and implemented constraint-based instruction tuning for reliable enterprise outputs.
Neurosymbolic AI Architect
AiqlNov, 2025 - Dec, 2025 1 month
Engaged as a Specialist Architect to design and validate the 'Digital Brain' blueprint for a Clinical Decision Support System. Delivered a functional Neurosymbolic GraphRAG prototype and a validated production roadmap. Architected a hybrid system using FastRP and GraphSAGE, conducted ablation studies, designed HITL feedback pipelines, validated GNNs via link split testing, engineered semantic grounding pipelines, deployed SHACL guardrails, and implemented automated audit layers for clinical risk discovery.
Founding Robotics Lead
Machines For HumansJan, 2025 - Oct, 2025 9 months
Served as Robotics Lead and Founding Engineer. Developed RL policies for teleoperation-augmented manipulation and unified locomotion/manipulation, led high-fidelity simulation design, integrated TensorRT-optimized models for real-time inference, engineered MLOps pipelines for autonomous systems, developed sim-to-real RL pipelines, and led a cross-functional team to deliver project milestones.
Senior Robotics Engineer
Kion GroupJul, 2024 - Jan, 2025 6 months
Led R&D in autonomous navigation, intelligent control systems, and machine learning-driven optimization for multi-agent autonomous mobile robots. Maintained the AMR navigation stack, architected a digital twin framework in NVIDIA Omniverse, implemented learning-based MPC for path planning, and oversaw software/hardware integration for robust robotic systems.
PhD Research Scholar/Assistant
Indian Institute Of Technology, GuwahatiJan, 2017 - Jan, 20247 yr
PhD research on bionic reflex control for underactuated robotic hand using reinforcement learning. Designed optimization algorithms for grasp quality, developed RL-based bionic reflex controllers, and robustified controllers using adaptive sliding mode control. Served as Teaching Assistant for AI, Robotics, and Mechatronics courses.
Achievements
- Gold Medal, Inter-Hostel Football Champion, Spardha, IIT Guwahati 2017
- Gold Medal, Inter-Hostel Football Champion, Spardha, IIT Guwahati 2018
- Silver Medal, Inter-Hostel Football Champion, Spardha, IIT Guwahati 2020
Major Projects
7Projects
Full-Stack LLMOps: Low-Resource NMT with Human Feedback (for English-Bodo pairs)
Architected a Transformer-based NMT model with supervised fine-tuning on a custom parallel corpus for English-Bodo. Designed and executed a full RLHF pipeline using Direct Preference Optimization, and deployed the model with Docker and FastAPI in a CI/CD pipeline.
Advanced End-to-End Graph Neural Network (GNN) Architectures for Knowledge Graph Reasoning
Benchmarked multiple GNN models including GraphSAGE and GAT on the ogbn-mag dataset. Deployed optimized GAT-based architecture to a robotics knowledge graph and built a production-ready REST API with Docker and FastAPI.
Neurosymbolic Reasoning for Explainable Robotic Affordance (AffordanceOWL)
Architected a hybrid neurosymbolic system integrating an OWL knowledge base with a Vision Language Model. Used a description logic reasoner for affordance inference and trained the VLM for improved zero-shot generalization.
Advanced Hybrid RAG Agent with Conversational Memory
Built a stateful, hybrid conversational agent using LangGraph, integrating a private RAG pipeline and real-time web search. Developed a two-model architecture with advanced prompt engineering and deployed a user-facing web interface with Streamlit.
Active Graph Repair: SME-Aligned Reasoning via DPO
Engineered a pipeline to extract physiological thresholds from medical guidelines into Neo4j, using RLHF with DPO for alignment. Implemented an active repair loop for HITL verification and achieved high resolution rates for numerical reasoning failures.
Neurosymbolic Causal Discovery for Supply Chain Resiliency
Architected a causal DAG in Neo4j using IOF/SCRO ontology for logistics disruption analysis. Integrated a Python-based causal engine for counterfactual reasoning and deployed a GNN for anomaly detection in supply chain data.
Enterprise Verified Agent Architecture: Neurosymbolic Multi-Agent Orchestration (Shadow Risk Detection)
Designed a privacy-preserving multi-agent platform decoupling neural and symbolic reasoning for financial workflows. Engineered predictive logic gates with Neo4j GDS and orchestrated a LangGraph squad for real-time explainability and CI/CD reliability.
Education
PhD (Robotics and AI)
Indian Institute of Technology, Guwahati (2024)M.Tech (Computer-Assisted Technology)
Indian Institute of Technology, Guwahati (2017)B.Tech (Mechanical Engineering)
National Institute of Technology, Silchar (2015)
Certifications
Reinforcement learning specialization
Deep learning specialization
Machine learning specialization
Mlops specialization
Natural language processing specialization
Generative adversarial networks(gans)
Llmops specialization
Deeplearning.ai tensorflow developer specialization
Tensorflow 2 for deep learning specialization
Prompt engineering
Bayesian statistics
Mathematics for machine learning specialization
Python, bash and sql essentials for data engineering
Postgresql for everybody specialization