Artificial Intelligence Developer

SAG Infotech Pvt Ltd; Jaipur; On-site; INR 1,200,000 - 2,200,000; Full time

Job summary

SAG Infotech Pvt Ltd in Jaipur is seeking a senior AI/ML engineer to design and deploy large language model fine-tuning pipelines, implement RLHF and instruction-tuning, and optimize inference efficiency. The role spans computer vision, multimodal systems, and autonomous AI agents integrated with external APIs and enterprise data.

You will work with PyTorch/TensorFlow, OpenCV, and modern MLOps stacks on cloud and edge deployment, ensuring scalable, reliable model serving and end-to-end AI

Qualifications

• 57 years of hands-on experience in data science, ML engineering, or applied AI research.
• Bachelor's or Master's degree in Computer Science, Statistics, Mathematics, or a related field (PhD is a plus).
• Deep proficiency in Python; strong command of PyTorch and/or TensorFlow/JAX.
• Proven experience fine-tuning transformer-based LLMs (open-source or proprietary) for downstream tasks.
• Solid background in computer vision model design, training, evaluation, and deployment.
• Experience building and deploying production-grade AI agents with tool use and memory systems.
• Strong backend development skills: RESTful APIs, microservices, cloud infrastructure (AWS/GCP/Azure).
• Proficiency with SQL and NoSQL databases; familiarity with vector databases for semantic search.
• Experience with containerisation (Docker, Kubernetes) and CI/CD pipelines for ML systems.
• Strong understanding of data structures, algorithms, and software engineering best practices.

Responsibilities

• Design and execute fine-tuning pipelines for large language models (GPT-4, LLaMA, Mistral, Gemma) using PEFT, LoRA, and QLoRA techniques.
• Implement RLHF, DPO, and instruction-tuning workflows to align model behaviour with domain-specific requirements.
• Optimize LLMs for inference efficiency using quantization (GPTQ, AWQ), distillation, and speculative decoding.
• Develop and deploy computer vision models for object detection, segmentation, OCR, and scene understanding using PyTorch and TensorFlow.
• Design, develop, and deploy advanced face recognition and biometric AI systems for face detection, verification, identification, clustering, and facial attribute analysis using PyTorch, TensorFlow, and OpenCV.
• Build high-performance face embedding pipelines using architectures such as FaceNet, ArcFace, InsightFace, RetinaFace, MTCNN, YOLO-based face detectors, and Vision Transformers (ViT).
• Develop anti-spoofing and liveness detection systems to prevent presentation attacks using RGB, IR, depth, and multimodal signals for secure biometric authentication.
• Build multimodal pipelines combining vision encoders (CLIP, ViT, SAM) with LLM decoders for visual question answering and image captioning.
• Apply video understanding techniques: action recognition, optical flow, and temporal modelling.
• Optimize vision models for real-time edge inference using TensorRT, ONNX, and CoreML.
• Design and build autonomous multi-agent systems using frameworks such as LangGraph, AutoGen, CrewAI, and custom orchestrators.
• Implement tool-use, function calling, memory management (short-term and long-term), and retrieval-augmented generation (RAG) for production agents.
• Develop agent evaluation frameworks to measure reliability, accuracy, and safety.
• Integrate agents with external APIs, databases, and enterprise systems via MCP and REST interfaces.

Skills

Python

PyTorch

TensorFlow

RESTful APIs

Kubernetes

Education

B.Tech/B.E.

MCA

M.Tech

Tools

Docker

OpenCV

YOLO

CLIP

LangChain

Job description

Key Responsibilities

LLM Fine-Tuning & Generative AI

• Design and execute fine-tuning pipelines for large language models (GPT-4, LLaMA, Mistral, Gemma) using PEFT, LoRA, and QLoRA techniques.
• Implement RLHF, DPO, and instruction-tuning workflows to align model behaviour with domain-specific requirements.
• Optimize LLMs for inference efficiency using quantization (GPTQ, AWQ), distillation, and speculative decoding.

Computer Vision & Multimodal Systems

• Develop and deploy computer vision models for object detection, segmentation, OCR, and scene understanding using PyTorch and TensorFlow.
• Design, develop, and deploy advanced face recognition and biometric AI systems for face detection, face verification, face identification, face clustering, and facial attribute analysis using PyTorch, TensorFlow, and OpenCV.
• Build high-performance face embedding pipelines using state-of-the-art architectures such as FaceNet, ArcFace, InsightFace, RetinaFace, MTCNN, YOLO-based face detectors, and Vision Transformers (ViT) for robust recognition under varying pose, illumination, occlusion, and low-resolution conditions.
• Develop anti-spoofing and liveness detection systems to prevent presentation attacks using RGB, IR, depth, and multimodal signals for secure biometric authentication.
• Build multimodal pipelines combining vision encoders (CLIP, ViT, SAM) with LLM decoders for visual question answering and image captioning.
• Apply video understanding techniques: action recognition, optical flow, and temporal modelling.
• Optimize vision models for real-time edge inference using TensorRT, ONNX, and CoreML.

AI Agent Development

• Design and build autonomous multi-agent systems using frameworks such as LangGraph, AutoGen, CrewAI, and custom orchestrators.
• Implement tool-use, function calling, memory management (short-term and long-term), and retrieval-augmented generation (RAG) for production agents.
• Develop agent evaluation frameworks to measure reliability, accuracy, and safety.
• Integrate agents with external APIs, databases, and enterprise systems via MCP and REST interfaces.

Backend Engineering & MLOps

• Build high-performance REST and gRPC APIs for model serving using FastAPI, Flask, or Django, with containerisation via Docker and Kubernetes.
• Design scalable ML pipelines using Apache Airflow, Prefect, or Kubeflow; manage data versioning with DVC and MLflow.
• Implement vector databases (Pinecone, Weaviate, Chroma, pgvector) for semantic search and RAG applications.
• Ensure model reliability with A/B testing, canary deployments, monitoring (Prometheus, Grafana), and drift detection.
• Collaborate with data engineers to build robust ETL/ELT pipelines on cloud platforms (AWS, GCP, Azure).

Required Qualifications

• 57 years of hands-on experience in data science, ML engineering, or applied AI research.
• Bachelor's or Master's degree in Computer Science, Statistics, Mathematics, or a related field (PhD is a plus).
• Deep proficiency in Python; strong command of PyTorch and/or TensorFlow/JAX.
• Proven experience fine-tuning transformer-based LLMs (open-source or proprietary) for downstream tasks.
• Solid background in computer vision model design, training, evaluation, and deployment.
• Experience building and deploying production-grade AI agents with tool use and memory systems.
• Strong backend development skills: RESTful APIs, microservices, cloud infrastructure (AWS/GCP/Azure).
• Proficiency with SQL and NoSQL databases; familiarity with vector databases for semantic search.
• Experience with containerisation (Docker, Kubernetes) and CI/CD pipelines for ML systems.
• Strong understanding of data structures, algorithms, and software engineering best practices.

Preferred Qualifications

• Experience with multimodal models (LLaVA, GPT-4V, Flamingo, Gemini) and vision-language tasks.
• Familiarity with model safety, red-teaming, and responsible AI evaluation methodologies.
• Publications in peer-reviewed venues (NeurIPS, ICML, CVPR, ICLR, ACL) or pre-prints on arXiv.
• Contributions to open-source ML projects or active Kaggle / HuggingFace community presence.
• Knowledge of efficient attention mechanisms (FlashAttention, Paged Attention) and serving frameworks (vLLM, TGI, Triton).
• Experience with streaming and real-time inference systems (Kafka, Redis Streams).
• Prior exposure to robotics, autonomous systems, or embodied AI is a bonus.

Core Technology Stack

Languages

Python, SQL, Bash; familiarity with Go or Rust a plus

ML Frameworks

PyTorch, TensorFlow, HuggingFace Transformers, Diffusers, PEFT

Computer Vision

OpenCV, YOLO, SAM, CLIP, ViT, TorchVision, MMDetection, Detectron2

LLMs & Agents

LangChain, LangGraph, AutoGen, CrewAI, OpenAI API, Anthropic API, Ollama, vLLM

Backend / Infra

FastAPI, Docker, Kubernetes, AWS (SageMaker, Lambda, S3), GCP Vertex AI

Data & MLOps

MLflow, Weights & Biases, Pinecone, Chroma, PostgreSQL, Redis, Kafka

Qualifications:

UG: B.Tech/B.E. in Any Specialization
PG: MCA in Any Specialization, M.Tech in Any Specialization