Senior AI/ML Engineer/ Manager

HSM Edifice Construction Services; Hingna; Sur place; INR 3 000 000 - 5 400 000; Plein temps

Résumé du poste

HSM Edifice Construction Services is seeking an AI Architect to own the technical design and delivery of an AI platform on AWS. This hands-on role translates business problems into production AI systems spanning generative AI, computer vision, and ML, setting architectural standards and roadmaps.

You will collaborate with product, data engineering, and MLOps teams to ensure scalable, secure, and cost-efficient solutions while guiding design reviews and mentoring engineers.

Qualifications

• 5+ years of professional experience in ML/AI, incl. architect/lead roles.
• Bachelor's or Master's in CS/Engineering/Math or equivalent practical experience.
• Strong Python; production experience with PyTorch and/or TensorFlow.
• Experience shipping generative AI systems to production (RAG, agents, fine-tuned LLMs).
• Solid CV background: CNNs, transformer-based vision models, training on real-world data.
• Deep AWS experience for ML workloads, incl. SageMaker + Bedrock/EKS/Lambda.
• Experience with MLOps tooling: Docker, Kubernetes, CI/CD, model versioning, and IaC.

Responsabilités

• Design end-to-end AI/ML architectures including data ingestion, feature stores, training, inference, monitoring, and feedback loops.
• Produce ADRs, reference designs, and roadmaps; lead design reviews with eng and security teams.
• Evaluate models/frameworks; own build-vs-buy and open-source vs managed decisions.
• Architect LLM-based solutions: RAG pipelines, agentic workflows, tool calling, and multi-step orchestration.
• Design CV systems for object detection, OCR, classification, segmentation, and video analytics; manage data pipelines.
• Architect AI workloads on AWS: SageMaker, Bedrock, Lambda, ECS/EKS, S3, Step Functions, API Gateway, Kinesis, OpenSearch.
• Mentor engineers; communicate trade-offs to stakeholders; contribute to hiring and long-term platform planning.

Connaissances

Python

PyTorch

TensorFlow

Generative AI

MLOps

Docker

Kubernetes

CI/CD

IaC

Architecture design

Communication

Formation

Bachelor's or Master's in Computer Science/Engineering/Math

Outils

MLflow

Weights & Biases

Terraform

CloudFormation

OpenSearch

Pinecone

Weaviate

LangChain

LlamaIndex

DSPy

vLLM

Triton

Description du poste

Job description

Role & responsibilities:

About the Role

We are looking for an AI Architect to own the technical design and delivery of our AI platform. This is a hands-on architecture role: you will translate business problems into production AI systems spanning generative AI, computer vision, and classical ML, and you will define how those systems are built, deployed, and scaled on AWS.

You will work across product, data engineering, and MLOps teams setting architectural standards, making build-vs-buy calls, and making sure what we ship is accurate, cost-efficient, secure, and maintainable.

Preferred candidate profile:

Key Responsibilities

Architecture & Design

• Design end-to-end AI/ML architectures covering data ingestion, feature stores, model training, inference serving, monitoring, and feedback loops.
• Produce architecture decision records (ADRs), reference designs, and technical roadmaps; drive design reviews with engineering and security stakeholders.
• Evaluate and select models, frameworks, and vendors; own build-vs-buy and open-source-vs-managed-service decisions.

Generative AI

• Architect LLM-based solutions: RAG pipelines, agentic workflows, tool/function calling, and multi-step orchestration.
• Design vector search and retrieval layers chunking strategy, embedding model selection, hybrid search, re-ranking.
• Implement fine-tuning, LoRA/PEFT, and prompt-optimization strategies where they beat prompting alone.
• Define evaluation frameworks for LLM output quality, hallucination rate, latency, and cost per request.
• Build in guardrails: prompt-injection defenses, PII redaction, content filtering, and human-in-the-loop review.

Computer Vision

• Design and deploy CV systems for [object detection / OCR & document extraction / image classification / segmentation / video analytics / quality inspection].
• Own the data pipeline for vision: annotation workflows, augmentation, class imbalance handling, and dataset versioning.
• Optimize models for the target environment quantization, pruning, distillation, ONNX/TensorRT conversion, and edge deployment where applicable.
• Work with multimodal models (vision-language) where they simplify or outperform traditional CV pipelines.

AWS & Platform

• Architect AI workloads on AWS using services such as SageMaker, Bedrock, Lambda, ECS/EKS, S3, Step Functions, API Gateway, Kinesis, and OpenSearch.
• Own cost architecture: instance/accelerator selection, spot strategy, autoscaling, caching, and token-spend controls.
• Implement CI/CD for models (MLOps): automated retraining, model registry, canary and blue-green rollouts, drift detection, and rollback.
• Ensure compliance with security, privacy, and data-residency requirements in partnership with the security team.

Leadership

• Mentor ML and software engineers; raise the bar through code and design review.
• Communicate technical trade-offs clearly to non-technical stakeholders and executives.
• Contribute to hiring, technical strategy, and long-term platform planning.

Required Qualifications:

• 5+ years of professional experience in machine learning, data science, or AI engineering, including 2+ years in an architect, lead, or senior IC role driving system design.
• Bachelor's or Master's in Computer Science, Engineering, Mathematics, or equivalent practical experience.
• Strong Python; production experience with PyTorch and/or TensorFlow.
• Demonstrated experience shipping generative AI systems to production RAG, agents, or fine-tuned LLMs not just prototypes or notebooks.
• Solid computer vision background: CNNs, transformer-based vision models, detection/segmentation architectures, and the practical realities of training on imperfect real-world data.
• Deep, hands-on AWS experience for ML workloads, including SageMaker and at least one of Bedrock, EKS, or Lambda-based inference.
• Experience with MLOps tooling: Docker, Kubernetes, CI/CD, model versioning (MLflow, Weights & Biases, or similar), and Infrastructure as Code (Terraform or CloudFormation).
• Proven ability to design for scale, latency, and cost you can reason about GPU utilization, throughput, and unit economics, not just accuracy metrics.
• Clear written and verbal communication; comfortable defending an architecture in front of a room.

Preferred Qualifications

• AWS certification (Machine Learning Specialty, or Solutions Architect Professional).
• Experience with vector databases (OpenSearch, Pinecone, Weaviate, pgvector, FAISS).
• Experience with LLM frameworks (LangChain, LlamaIndex, DSPy) and serving stacks (vLLM, TGI, Triton).
• Edge or real-time CV deployment (NVIDIA Jetson, GStreamer, RTSP pipelines).
• Distributed training experience (DeepSpeed, FSDP, Ray).
• Streaming and big-data experience (Kafka, Spark).
• Familiarity with responsible AI practice: bias evaluation, model cards, explainability, and emerging regulation (EU AI Act, NIST AI RMF).
• Domain exposure in [healthcare / manufacturing / retail / fintech / logistics].
• Open-source contributions, patents, or published research