Job Overview

Roles & Responsibilities

• Deep Learning 
• Architect and train CNN/ViT models for classification, detection, segmentation,
• and OCR.
• Build and optimize RNN/LSTM/GRU models for sequence learning, speech, or timeseries
• forecasting.
• Research and implement transformer-based architectures bridging vision and
• language tasks.
• Create scalable pipelines for data ingestion, annotation, augmentation, and
• synthetic data generation.
• Agentic AI & Multi-Agent Frameworks
• Design and implement multi-agent workflows using LangChain, LangGraph, CrewAI,
• or similar frameworks.
• Develop role hierarchies, state graphs, and integrations that enable autonomous
• vision + language workflows.
• Optimize agent systems for latency, cost, and reliability.
• LLM Fine-Tuning & Retrieval-Augmented Generation (RAG)
• Fine-tune open-weight LLMs using LoRA/QLoRA, PEFT, or RLHF methods.
• Develop RAG pipelines integrating vector databases (FAISS, Weaviate, pgvector).
• Combine LLM reasoning with CNN/RNN perception modules in multimodal systems.
• MLOps & Deployment at Scale
• Develop reproducible training workflows with PyTorch/TensorFlow and experiment
• tracking (W&B, MLflow).
• Deploy models with TorchServe, Triton, or KServe on cloud AI stacks (AWS
• Sagemaker, GCP Vertex, Kubernetes).
• Optimize inference with ONNX/TensorRT, quantization, and pruning for cloud and edge devices.
• Build robust APIs/micro-services (FastAPI, gRPC) and ensure CI/CD, monitoring,
• and automated retraining.

Desired Candidate

• B.S./M.S. in Computer Science, Electrical Engineering, Applied Math, or related
• discipline.
• 5+ years building deep learning systems with CNNs and RNNs in production.
• Strong Python skills and Git workflows.
• Proven delivery of computer vision pipelines (OCR, classification, detection).
• Hands-on experience with LLM fine-tuning and multimodal AI.
• Experience in containerization (Docker) and deployment on cloud AI platforms.
• Knowledge of distributed training, GPU acceleration, and inference
• optimization.
• Preferred Qualifications
• Research experience in transformer architectures (ViTs, hybrid CNN-RNNTransformer
• models).
• Prior work in sequence modeling for speech or time-series data.
• Contributions to open-source deep learning frameworks or vision/sequence
• datasets.
• Experience with edge AI deployment and hardware optimization

Competencies