I build intelligence that moves beyond the screen.
I'm Krishna Teja, an ML and robotics engineer at Accenture in Munich. I connect perception, learning and control so that a model ends up as a robot doing something useful.
I started in automotive perception at ESR Labs. Now I work on manipulation, humanoid teleoperation and making models run fast on real hardware.
Move your pointer over the cell. The arm solves inverse kinematics every frame.
- θ₁
- 0.0°
- θ₂
- 0.0°
- x
- 0 mm
- y
- 0 mm
What I believe about building robots
From my notesA pose is only useful if you can trust it.
Before a model goes on a robot, measure it on your object, your camera and your lighting. Benchmarks don't tell you how it fails in your cell.
Write the robot layer once.
A clean, robot-agnostic interface turned a Franka-to-UR5e port into a three-day job instead of a rewrite.
Document the limitations, not just the wins.
Every write-up ends with what broke. The next iteration starts with better questions.
Latency is a design choice.
Gloves and trackers over camera hand-tracking, TensorRT over raw PyTorch. Compute, memory and throughput come up in almost every project.
Path so far
- 2016 – 2017MSc Computer Science · Kaiserslautern
Master thesis at Audi Electronics Venture: anticipating lane changes and turns from driver video and CAN data, ~90% accuracy.
- 2017 – 2023Automotive perception · ESR Labs
Segmentation, detection, depth and rain detection, optimised and deployed on Drive PX and other embedded targets.
- 2023 – 2025Pose estimation & teleoperation
OnePose++, FoundationPose evaluation on a Franka FR3, hand-eye calibration, a first glove prototype.
- Now · AccentureManipulation & humanoids
LLM-driven cobots, lab automation on a UR5e, humanoid teleop from Isaac Lab to a real Unitree G1.
Toolbox
6D pose · FoundationPose · CNOS · LoFTR · PyTorch · TensorRT
ROS 2 · Isaac Lab · Docker · hand–eye calibration · C++
Franka FR3 · UR5e · Viper 300s · Unitree G1 · RealSense · Zivid · SenseGlove