Technical Skills
Practical experience in robot manipulation, simulation, control, state-aware planning, robot evaluation, and simulation-to-real deployment.
Core profile:
Robotics engineer with hands-on experience building simulation
environments, Cartesian controllers, manipulation pipelines,
natural-language robot agents, evaluation infrastructure, and
real Franka robot applications.
Programming & Software Development
- Python development for robotics, simulation, data processing, APIs, planning, and automation
- C++ and C for robotics, embedded systems, and performance-oriented software
- C# experience from application and engineering projects
- Object-oriented programming, modular software design, debugging, and testable workflows
- JSON, CSV, YAML, command-line tools, and structured experiment artifacts
Python
C++
C
C#
Bash
VHDL
HTML / CSS
Robotics Simulation
- Building manipulation environments and robot tasks in NVIDIA Isaac Sim and MuJoCo
- ROS 2 integration with simulation, command interfaces, navigation, and robot-control workflows
- Gazebo and Webots simulation for mobile robotics, mapping, navigation, and obstacle handling
- Robot-scene modeling, object placement, cameras, collision properties, contact behavior, and physics debugging
- Simulation experiment design and persistent scene-state management
NVIDIA Isaac Sim
MuJoCo
Gazebo
Webots
ROS 2
RViz
URDF / USD
Robot Manipulation & Control
- Cartesian translation and rotational control for Franka Panda and FR3 robots
- Inverse kinematics, Jacobian-based control, end-effector pose control, and trajectory execution
- Pick, grasp, lift, transport, place, stack, and release manipulation primitives
- Contact-rich manipulation, screw insertion, force observation, and safety-limit handling
- Teleoperation and real-time robot-feedback interfaces
Franka Panda
Franka FR3
Cartesian Control
Inverse Kinematics
Jacobians
Grasping
Contact-Rich Tasks
Planning, Scene Reasoning & Robot Agents
- Natural-language command interfaces for robot manipulation
- Scene-graph reasoning for support relationships and object-blocker detection
- State-aware planning for stacked objects, occupied destinations, and temporary buffer movements
- Procedural geometric task generation for previously unseen letters and digits
- Safety validation for robot workspace boundaries, target spacing, and approved manipulation primitives
Scene Graphs
Task Planning
State-Aware Planning
Qwen3
Ollama
Streamlit
Geometric Planning
Robot Learning & Embodied AI
- Language-conditioned manipulation and symbolic action representations
- Behavior-cloning data preparation and demonstration replay
- Failure-driven scenario generation for imitation learning, reinforcement learning, and VLA evaluation
- Action encoding, teacher-student correction, residual policy experiments, and policy comparisons
- Introductory implementations of diffusion-style action prediction, world models, and preference learning
VLA
Behavior Cloning
Imitation Learning
Reinforcement Learning
Action Encoding
World Models
Policy Evaluation
Failure Analysis & Robot Evaluation
- Episode-level evaluation using success, latency, placement error, and safety metrics
- Baseline-versus-candidate comparisons and regression detection
- Automated SHIP / BLOCK release-gate decisions
- Failure-report generation and targeted retry-scenario creation
- Reproducible rollout artifacts including metrics, reports, JSON logs, and videos
ContactTrace AI
Regression Gating
Failure Diagnosis
Metrics
CI/CD
SHIP / BLOCK
Computer Vision & Sensor Processing
- RGB and RGB-D camera processing for object localization and manipulation targets
- Color segmentation, image masks, pixel-to-world back-projection, and depth processing
- OpenCV and MediaPipe for visual and gesture-based interfaces
- EMG and IMU preprocessing, filtering, feature extraction, and gesture classification
OpenCV
MediaPipe
RGB-D
Depth Processing
EMG
IMU
Signal Filtering
Machine Learning & Data Analysis
- Classical machine-learning models including SVM, KNN, logistic regression, random forests, and GMM
- Model training, validation, confusion analysis, and feature engineering
- NumPy and Pandas for experiment processing and dataset preparation
- PyTorch fundamentals and neural-network prototyping
scikit-learn
NumPy
Pandas
PyTorch
SVM
Random Forest
Feature Engineering
Backend, Infrastructure & DevOps
- FastAPI services for robotics evaluation and experiment management
- Asynchronous task processing with Celery and Redis
- PostgreSQL databases and MinIO-compatible artifact storage
- Docker-based development and reproducible service deployment
- GitHub Actions and CI-style robot regression checks
FastAPI
Celery
PostgreSQL
Redis
MinIO
Docker
GitHub Actions
Development Tools & Workflow
- Linux development, Bash workflows, Python environments, and dependency management
- Git version control, GitHub repositories, releases, and collaborative workflows
- VS Code, Visual Studio, terminal-based debugging, and log analysis
- Structured experiment logging and reproducible project organization
Ubuntu Linux
Git
GitHub
VS Code
Conda
venv
Bash
Embedded Systems & Engineering Tools
- Arduino and microcontroller-based sensing and actuation
- Sensor integration, timers, relays, displays, and hardware–software interfacing
- Foundational experience with circuit simulation, digital design, and electronics tools
Arduino
Microcontrollers
Raspberry Pi
Cadence Virtuoso
PSpice
Proteus
AutoCAD
Languages
- English: Fluent professional proficiency
- German: A2 level, currently preparing for B1
- Bengali: Native
Engineering & Professional Strengths
- Structured debugging of robotics, simulation, control, and integration problems
- Translating research concepts into working technical prototypes
- Metrics-driven experimentation and evidence-based engineering decisions
- Clear technical documentation, presentations, and cross-functional communication
- Independent project ownership, rapid learning, and iterative development
Problem Solving
Technical Communication
Research Prototyping
Experiment Design
Documentation
Project Ownership