artificial intelligence
Our Direction
Core AI / ML Foundations
- Focus: Explainable, robust, and automotive-grade intelligence for diagnostics, monitoring, and decision logic.
Technical competencies
- Supervised Learning - Random Forest, Adaptive Random Forest, XGBoost, LightGBM, SVM, k-NN, Logistic Regression
- Reinforcement Learning- PPO, DQN, Double DQN, A2C/A3C, SAC, Q-Learning, MDP
- Statistical Monitoring & Anomaly Detection- EWMA, CUSUM, Bayesian Change Detection, GMM, Isolation Forest, One-Class SVM, Kalman Filter
- Deep Learning - ANN, CNN, RNN, LSTM, GRU, TCN, Autoencoders, Transformers
- Edge AI & Embedded AI- TensorFlow Lite, ONNX Runtime, TensorRT, Quantization, Pruning, ARM Cortex, NVIDIA Jetson
- Time-Series & Signal Processing- FFT, STFT, Wavelets, ARIMA/SARIMA, Kalman Filter, Butterworth Filters, Feature Extraction
- Language & Text AI- BERT, GPT, LLMs, RAG, Sentence Transformers, FAISS, Prompt Engineering
- Agentic AI Systems- LangChain, LangGraph, AutoGen, CrewAI, ReAct, Tree-of-Thoughts, MCP
- Multimodal AI- CLIP, Multimodal Transformers, Sensor Fusion, Cross-Modal Learning, GNNs
- Multimodal AI- CLIP, Multimodal Transformers, Sensor Fusion, Cross-Modal Learning, GNNs
- MLOps & AI Lifecycle- MLflow, Kubeflow, DVC, Jenkins, GitHub Actions, Kafka, Airflow
Engineering value:
- Reliable fault detection and behavioral analysis
- Adaptive decision logic under dynamic conditions
- AI models suitable for real-time and safety-aware environments
Language & Agentic AI Systems
- Focus: Knowledge-driven automation and autonomous reasoning for engineering and in-vehicle intelligence.
Technical competencies
- Language & Text AI: LLM-based code and document synthesis, domain-grounded reasoning using RAG
- Agentic AI: Goal-driven, tool-using AI systems (LangChain-based), multi-agent collaboration, human-in-the- loop control
- Multimodal Intelligence: Text, code, and structured data understanding and generation
Engineering value:
- Reliable fault detection and behavioral analysis
- Adaptive decision logic under dynamic conditions
- Real-time and safety-aware AI model deployment
- Accelerated development and validation workflows
- Context-aware reasoning and decision support
- Intelligent infotainment, diagnostics, and HMI enablement
- Integration into embedded and automotive-grade systems
- Support for predictive maintenance and system optimization
current ready mini projects
AI MVP (Minimum viable products)
1.Agentic AI for Requirement Analysis-
AI-driven extraction, validation, and Q&A across specifications, requirements, and organizational knowledge bases for faster and consistent requirement engineering.
2.Agentic AI for Software Design-
Automated generation of UML artifacts including activity, class, and sequence diagrams to accelerate architecture definition and design consistency.
3.Agentic AI for Code Configuration & Generation-
Intelligent automation of AUTOSAR configuration and code generation across ARXML, XDM, MATLAB, Mbed OS, STM32Cube, Git, and VS Code ecosystems.
4.Agentic AI for Component & Integration Testing-
AI-powered test generation and execution ensuring statement, branch, and MC/DC coverage with end-to-end requirement traceability.
5.Agentic AI for Software Qualification Testing-
Automated validation frameworks leveraging Python, CAPL, and integrated test tools for scalable, repeatable, and standards-compliant testing.
6.Agentic AI for SDV Diagnostics & Self-Healing Systems-
End-to-end intelligent diagnostics enabling real-time fault detection, root cause analysis, and autonomous self-healing in Software-Defined Vehicles (SDVs).