Download full PDF resume · Consulting offerings
Summary
Senior ML engineer and AI strategy consultant with 6+ years across ADAS perception, aerial segmentation, and edge robotics. Advises on AI fit and integration—readiness assessments, MLOps audits, and practical roadmaps for ML/Gen AI when the business case is there—alongside hands-on delivery in computer vision and 3D perception (TensorRT/Jetson), production MLOps (CI/CD, experiment tracking, monitoring), and private Gen AI (RAG, on-prem deployment). Leads work end to end as a contractor and tech lead, from synthetic data and sim-to-real through latency budgets and stakeholder-ready outcomes.
Professional Experience
Sr. ML Research Engineer
Verses AI · Remote (Contractor) · Dec 2023 – Dec 2025
- Edge 3D perception for warehouse detection and depth-aware navigation on Jetson-class hardware (YOLOv10n, BlenderProc2 synthetic data, TensorRT).
- Improved detection mAP (0.15 → 0.62 synthetic; 0.58 real) and cut end-to-end latency to 26 ms at 640×640; combined depth + detection at 10 FPS on Jetson Orin NX.
ML Tech Lead
Verses Global BV (dAIEdge) · Remote (Contractor) · Dec 2023 – Mar 2026
- Active Inference routing agent on the perception stack; JAX planning optimized from 7 min → 21 s.
- PyMDP saccading agent on an IoT camera; Isaac Omniverse simulation for warehouse routing evaluation.
- Publication: Towards smart and adaptive agents for active sensing on edge devices (HiPeac 2025).
AI Strategy Consultant
Dev AI Ventures (Self-Employed) · Remote · Jan 2024 – Present
- Docai.tools: OCR → private RAG architecture (chunking, embeddings, deployment).
- SOTA VLM research for Drizz.dev; deep-learning strategy for Eventgraphia (image culling).
Machine Learning Engineer
Edge Case Research GmbH · Munich · May 2020 – Jan 2022
- End-to-end ADAS perception with MOT on AWS; 2D detection extended to 3D bounding boxes.
- Production monitoring and alerting with Weights & Biases across model versions.
Machine Learning Engineer
TerraLoupe GmbH · Munich · Oct 2019 – Apr 2020
- Docker/CI/CD MLOps for multi-TB aerial segmentation; Sacred/Omniboard across 20+ iterations (~40% faster experimentation).
- GPU vs TPU benchmarking to inform production infrastructure strategy.
ADAS Engineer
KPIT Technologies GmbH · Munich · Jan 2019 – Aug 2019
- Vehicle State Monitor for 100+ Hz multi-sensor telemetry; real-time health GUI and automated test framework.
Software Engineer
CNRS (XLIM Lab) · Poitiers · Jan 2017 – Aug 2018
- Radio propagation simulator optimization (~30% runtime reduction).
- Publication: CupCarbon (CCNC 2018).
Research Experience
Fraunhofer AISEC
Munich · Feb 2022 – Sep 2023
- Robustness/privacy research on CNNs, Lipschitz bounds, and DP threats; containerized reproducible evaluation pipelines.
- Preprint: Gradient masking and underestimated robustness threats of DP in deep learning.
Master’s Thesis · Synthetic Adversarial Generation
Fraunhofer AISEC · Feb – Aug 2023 · Grade 1.0/5.0
- DDPM synthetic adversarial datasets for CIFAR-10 robustness retraining (+8–10% ResNet18 accuracy).
Warp Automotive / UnternehmerTUM
Munich · Nov 2020 – Mar 2021
- Urban pedestrian segmentation and LiDAR PoC; MLOps pipeline (Docker, experiment tracking, Go-task automation).
CAMP Chair, TUM
Munich · Apr – Jul 2020
- U-Net multi-coil MRI 3D reconstruction (3rd place globally); NeRF exploration for CT-from-X-ray.
- Publication: Multi-Coil MRI Reconstruction Challenge (Frontiers in Neuroscience).
Education
M.Sc. Informatics (ML & CV)
Technical University of Munich · Oct 2019 – Aug 2023
B.Tech Computer Science
LNM Institute of Information Technology, Jaipur · Jul 2012 – Jun 2016
Technical Skills
Gen AI: vLLM, Ollama, RAG pipelines, LoRA, agent workflows ·
MLOps: Docker, FastAPI, W&B/MLflow, CI/CD (GitHub Actions, Jenkins), Kubernetes ·
Cloud: AWS, GCE ·
Core: PyTorch, JAX, TensorFlow, OpenCV, ROS/ROS2 ·
Languages: Python, C++, Java, MATLAB, bash
Contact
- Email: vyasdevms@gmail.com
- LinkedIn: devendravyas
- GitHub: skat00sh
- Publications: Google Scholar