AI Engineer

We re seeking an AI Engineer with strong hands-on experience in computer vision and multimodal (vision + voice) systems from data preparation and model training to scalable inference and production deployment. You ll build and ship end-to-end AI solutions: curate datasets, fine-tune and evaluate models, optimize inference pipelines, and operationalize services on cloud infrastructure. Job Requirements Bachelor s degree in Computer Engineering / Software Engineering / Electronics / Computer Science or equivalent (Master s preferred). 0-3 years building production AI systems, with a focus on computer vision (PyTorch/TensorFlow). Strong Python engineering skills; experience with REST/JSON APIs (FastAPI/Flask), and service design. Hands-on with cloud deployment for AI (AWS/GCP/Azure), containers (Docker), orchestration (Kubernetes/Cloud Run/ECS), and GPUs. Experience training and serving CV models (ResNet/EfficientNet, YOLO/RetinaNet, U-Net/nnU-Net, Vision Transformers), and familiarity with ASR/TTS pipelines (e.g., Whisper, torchaudio). Practical knowledge of model optimization (ONNX/TensorRT, quantization), and data tooling (Pandas, NumPy, OpenCV, ffmpeg). Proficiency with MLOps tooling: experiment tracking (MLflow/W&B), model registry, CI/CD, monitoring/logging (Prometheus/Grafana/Cloud Monitoring). Strong grounding in evaluation methodology: metrics, ablation studies, error/bias analysis, and reproducible research practices. Comfort with cloud storage, queues, and databases; experience integrating AI services into existing systems. Computer vision: 2 years (Preferred) Nice-to-Have: Experience with multimodal (vision + workflows; Hugging Face ecosystem. Knowledge of real-time/edge inference, TensorRT-LLM, vLLM. Background in signal processing or audio engineering; speaker diarization, voice cloning ethics. Experience in regulated domains with HITL workflows and documentation. Main Job Duties Model development and inference Implement, fine-tune, and evaluate CV models (classification, detection, segmentation, tracking) using PyTorch/TensorFlow. Build robust inference services (REST/JSON) with batching, streaming, and hardware acceleration (GPU, TensorRT/ONNX). Train and adapt voice/ASR/TTS models and integrate with vision pipelines (multimodal workflows). Data and evaluation Own dataset lifecycle: collection, cleaning, labeling/specs, augmentation, and versioning. Define metrics and test sets; run offline/online evaluations (accuracy, latency, throughput, calibration) and error analysis. Develop data transformation and feature pipelines; maintain data quality checks and bias/fairness assessments. Production engineering Containerize and deploy models to cloud (AWS/GCP/Azure) using Docker/Kubernetes/Cloud Run/ECS. Implement CI/CD, experiment tracking, model registry, A/B canaries, rollout/rollback strategies. Build monitoring for drift, performance, and cost; automate retraining or active learning loops. Systems integration Design APIs and modules, integrate with upstream/downstream systems, and ensure reliable contracts and observability. Collaborate with product, design, and backend teams to turn requirements into measurable deliverables. Performance and optimization Optimize training/inference (quantization, pruning, distillation, mixed precision); leverage ONNX/TensorRT/torch.compile. Profile and tune data loaders, GPU utilization, caching, and I/O. Compliance and safety Implement data governance, privacy, and security best practices; maintain audit trails and documentation. Establish HITL workflows and guardrails for clinical or safety-critical contexts where applicable.