AI/ML Engineer

Jash Data Sciences Pvt. Ltd.; Pune District; On-site; INR 900,000 - 1,500,000; Full time

Job summary

Jash Data Sciences Pvt. Ltd. in Pune, India, seeks an AI-ML Engineer to join a fast-growing startup. You will design, build, and deploy production-grade ML, DL, and Generative AI applications, bridging software engineering with cutting-edge AI research.

The role emphasizes scalable architectures, MLOps, and high-throughput inference with exposure to LangChain, vector databases, and distributed training. Strong Python skills, DL expertise, and responsible AI are essential.

Qualifications

• Strong Python programming skills with hands-on project experience, Git proficiency, and a basic understanding of CUDA.
• Expertise in Deep Learning architectures (Transformers, CNNs, RNNs) with strong ML theory.
• Deep proficiency in PyTorch or TensorFlow with custom layer implementation and training loops.
• Experience with NLP, Computer Vision, or Time Series Analysis.
• Proven experience writing clean, production-grade code with pytest and Great Expectations.

Responsibilities

• Deliver end-to-end AI engineering projects by designing, building, and deploying production-grade ML, DL, and GenAI applications.
• Develop high-quality software solutions in Python, integrating AI models into application codebases.
• Implement advanced training strategies including mixed-precision training and distributed training with GPU utilization.
• Optimize, serialize (ONNX, TorchScript), and deploy models via REST APIs (FastAPI).
• Implement robust MLOps workflows using DVC, Docker, and cloud platforms for model versioning and monitoring.
• Architect and optimize Retrieval-Augmented Generation systems using hybrid search and cross-encoder reranking.
• Design autonomous, multi-agent workflows using function calling and ReAct patterns.
• Implement serving techniques (vLLM, quantization) to maximize throughput.
• Move notebooks to production-grade pipelines with unit/integration tests (pytest).
• Diagnose production anomalies and training bottlenecks.
• Evaluate LLM outputs with appropriate metrics and tests.
• Design high-throughput data pipelines and optimized SQL/NoSQL queries for data processing.

Skills

Python programming

Git proficiency

CUDA basics

Deep Learning (Transformers)

PyTorch/TensorFlow

NLP/Computer Vision/Time Series

Testing with pytest/Great Expectations

Education

Bachelor's or Master's degree in CS/AI/DS

Tools

ONNX

TorchScript

FastAPI

Docker

DVC

MLflow

Pinecone/Chroma/Weaviate

Job description

We are a fast-growing startup based in Pune, India, specializing in cutting-edge Data Science and Data Engineering solutions. Our team of dedicated professionals is committed to solving complex data challenges for companies worldwide.

Our Culture

We foster a vibrant startup culture that values:

• Continuous learning
• Positive work environment
• Collaborative problem-solving

Role Overview

We are seeking a highly technical and proactive AI-ML Engineer to join our dynamic team. The ideal candidate will possess a strong blend of software engineering rigor and deep technical expertise in modern AI/ML architectures, infrastructure optimization, and generative AI systems. This role demands critical thinking, production-grade coding, and the ability to deploy, optimize, and scale robust AI models to solve complex, real-world problems. You should bring a deep curiosity for understanding model internals, and the adaptability required to navigate and implement rapidly evolving, state-of-the-art technologies.

Key Responsibilities

• Deliver end-to-end AI engineering projects by designing, building, and deploying production-grade Machine Learning, Deep Learning, and Generative AI applications.
• Develop high-quality software solutions in Python, collaborating with cross-functional engineering teams to integrate AI models into existing application codebases.
• Implement advanced training strategies including mixed-precision training (FP16/BF16), gradient accumulation, and distributed training (Data/Model/Pipeline parallel) while profiling and maximizing GPU utilization.
• Optimize, serialize (ONNX, TorchScript), and deploy models via high-throughput REST APIs (FastAPI) while managing latency vs. throughput trade-offs.
• Implement robust MLOps workflows using DVC, Docker, and cloud platforms for model versioning, pipeline automation, and production monitoring.
• Architect and optimize high-performance Retrieval-Augmented Generation (RAG) systems using hybrid search, semantic chunking, and cross-encoder reranking.
• Design autonomous, multi-agent systems and orchestrated workflows utilizing function calling, the ReAct pattern, and advanced memory management architectures.
• Implement state-of-the-art serving techniques (vLLM, speculative decoding, prompt caching) and quantization (INT8/INT4 via GPTQ/AWQ) to maximize inference throughput.
• Move workflows from notebooks to production-grade pipelines, writing clean code and implementing unit/integration tests for ML (pytest, Great Expectations).
• Actively diagnose production anomalies, including data/concept distribution shifts, silent model failures, and training bottlenecks (underfitting/overfitting).
• Evaluate and benchmark LLM outputs using appropriate metrics and testing frameworks.
• Design high-throughput data pipelines and optimized SQL/NoSQL queries for large-scale data processing and model feature injection.
• Practice active listening to understand project requirements and team inputs.
• Collaborate with stakeholders to translate complex business requirements into scalable AI/ML solutions and communicate technical trade-offs clearly.
• Demonstrate strong technical communication skills, a high degree of ownership, and an action-biased approach to debugging and solving ambiguous engineering problems
• Apply responsible AI principles, jailbreak awareness, and output validation guardrails to ensure ethical and safe model development.
• Plan strategically and multitask efficiently to meet project deadlines.

Required Skills

Core Programming & ML

• Strong Python programming skills with hands-on project experience, Git proficiency, and a basic understanding of CUDA
• Expertise in Deep Learning architectures (Transformers, CNNs, RNNs) alongside a strong theoretical understanding of foundational ML algorithms (GBMs, Random Forests)
• Deep proficiency in PyTorch or TensorFlow, with a strong emphasis on custom layer implementation and neural network training loops
• Hands-on experience with modern training paradigms, including Self-supervised Learning, Contrastive Learning, and advanced Transfer Learning
• Experience with NLP, Computer Vision, or Time Series Analysis
• Proven experience writing clean, production-grade code, utilizing pytest and Great Expectations for data and model validation

Generative AI & LLMs

• Hands-on experience with commercial LLM APIs (OpenAI, Anthropic, Groq) and hosting/deploying open-source foundational models (Llama, Mistral)
• Proficiency with modern orchestration frameworks (LangChain, LlamaIndex) and building stateful multi-agent architectures using LangGraph or DSPy
• Experience with vector databases (Pinecone, Weaviate, Chroma, pgvector), advanced chunking (fixed, semantic, recursive), and hybrid search implementation
• Deep technical understanding of Parameter-Efficient Fine-Tuning (PEFT) mechanics - specifically LoRA/QLoRA low-rank decomposition - alongside instruction tuning and model alignment methodologies (DPO, RLHF)
• Mastery of advanced prompt engineering, including structured output forcing, Chain-of-Thought, system prompt design, and building resilient multi-agent coordination systems
• Hands-on experience with inference optimization and high-throughput serving, including quantization (GPTQ, AWQ), speculative decoding, prompt caching, and vLLM acceleration

MLOps & Deployment

• Experience with MLOps practices, logging, and model registries (MLflow, Weights & Biases, DVC) along with model serving via Triton Inference Server or FastAPI
• Production experience deploying, scaling, and monitoring models natively on cloud AI platforms, specifically AWS SageMaker or GCP Vertex AI
• Experience building CI/CD pipelines for ML applications, with an emphasis on data and model versioning/lineage using DVC and Git
• Solid understanding of SQL, including advanced concepts like windowing functions and query optimization
• Experience building and orchestrating data pipelines using Airflow or Prefect to feed specialized SQL and NoSQL vector/feature databases

Soft Skills & Professional Attributes

• Strong critical thinking and problem-solving skills
• Excellent written and verbal communication abilities
• High degree of flexibility and adaptability to stay ahead of the rapid velocity of the open-source AI ecosystem (Hugging Face, vLLM, etc.)
• Practical understanding of AI safety, compliance, guardrails (e.g., NeMo Guardrails or Llama Guard), and responsible AI practices.

Nice-to-Have

• Contributions to open-source AI/ML repositories or GenAI orchestration frameworks.
• Experience with real-time streaming data processing
• Active participation in competitive ML spaces (Kaggle) or track record of reviewing/reproducing state-of-the-art AI research papers
• Published research papers or conference presentations
• Experience building and scaling Graph Databases or Knowledge Graphs for advanced RAG
• Experience with multimodal architectures (Vision-Language models, Audio processing)

Qualifications

• AI-ML Engineer: 2-5 years of hands-on experience engineering machine learning systems, optimizing infrastructure, and implementing LLM/GenAI workflows in production
• Bachelor's or Master's degree in Computer Science, Artificial Intelligence, Data Science, Statistics, or a related highly quantitative field
• Demonstrated commitment to continuous learning through contributing to open-source, certifications, or self-study (especially in deep learning internals, MLOps, and modern GenAI frameworks)

What We Offer

• Competitive salary commensurate with experience
• Opportunity to work on diverse, cutting-edge AI/ML projects
• Collaborative and innovation-driven work environment
• Rapid growth and continuous learning opportunities
• Exposure to latest AI technologies and industry best practices