Data Scientist

GK HR Consulting India Private Limited; Bengaluru; Presencial; INR 2.500.000 - 4.000.000; Jornada completa

Descripción de la vacante

GK HR Consulting India Private Limited is seeking a Data Scientist with 5+ years of hands-on experience in RAG architectures, LLMs, and enterprise-grade AI solutions in Bengaluru. You will work on grounding responses using enterprise data, and design multi-step agent workflows for controlled execution across IoT and automation use cases.

Responsibilities include enhancing ML/DL pipelines, building scalable model training and deployment pipelines, and defining evaluation metrics for grounding,

Formación

• Master’s degree in CS/EE/Math or related field; PhD preferred.
• Strong ability to translate ambiguous objectives into flexible AI solutions.
• Proven track record delivering impactful AI outcomes in complex environments.

Responsabilidades

• Analyze existing digital products to understand intelligent models and improve performance and scalability.
• Enhance ML/DL pipelines with LLM-based capabilities such as summarization, Q&A, reasoning, and copilots.
• Design and implement LLM-based RAG solutions grounding responses on enterprise data.
• Build agentic AI workflows enabling multi-step task planning, tool invocation, and guarded execution.
• Develop agent orchestration patterns and fallback mechanisms for low-confidence retrievals.
• Drive innovation through experimentation, inventions, and novel solution approaches.

Conocimientos

RAG architecture

LLM integration

API integration

Agent workflows

R&D applications

Model evaluation

Deployment patterns

Educación

Master's degree

PhD preferred

Herramientas

Pinecone/Milvus/Weaviate

LangChain

Kubernetes

Docker

Vector databases

Descripción del empleo

Data Scientist-5+ years, RAG architecture, API integrations, LLM Ops tools, R& D applications

Key Responsibilities

• Analyze existing digital products to understand current intelligent models and improve their performance, reliability, and scalability.
• Enhance traditional ML and DL pipelines by incorporating LLM-based capabilities such as summarization, Q&A, reasoning, decision support, and copilots.
• Design and implement LLM-based solutions using Retrieval‑Augmented Generation (RAG) to ground responses on enterprise data including documents, manuals, telemetry, tickets, and knowledge bases.
• Build Agentic AI workflows that enable multi-step task planning, tool and API invocation through function calling, controlled action execution with guardrails and approvals, and contextual memory management.
• Develop agent orchestration patterns such as multi-agent collaboration (planner-executor-critic), deterministic workflow engines, and fallback mechanisms for low-confidence retrieval or reasoning.
• Drive innovation through experimentation and contribute to invention disclosures, patents, and novel solution approaches.
• Design and implement AI solutions for IoT, robotics, and automation use cases.
• Build and maintain scalable pipelines for model training, evaluation, and deployment across batch and real-time inference scenarios.
• Manage experiment tracking, model versioning, and model registries to ensure reproducibility, traceability, and governance.
• Define and track LLM-specific evaluation metrics, including groundedness, faithfulness, hallucination rate, toxicity, and safety.
• Monitor retrieval system quality using metrics such as precision, recall, chunking effectiveness, latency, and knowledge coverage.

Required Qualifications

• Master’s degree in Computer Science, Electrical Engineering, Applied Mathematics, Statistics, or a related field (PhD preferred).
• Strong oral and written communication skills; ability to explain technical concepts to non-technical stakeholders.
• Demonstrated ability to take ambiguous objectives and design innovative, flexible solutions.
• Proven track record of delivering impactful outcomes and driving change in complex environments.

Required Technical Skills

• Strong expertise in Large Language Models (LLMs) and building scalable, production-grade applications using them.
• Hands-on experience designing and implementing Retrieval‑Augmented Generation (RAG) architectures.
• Experience building document ingestion and preprocessing pipelines for unstructured and semi-structured data.
• Expertise in defining effective chunking strategies to optimize retrieval quality and context relevance.
• Strong understanding of embeddings, vector representations, and vector search techniques.
• Experience implementing retrieval and reranking mechanisms to improve response accuracy.
• Familiarity with grounding and citation strategies to ensure reliable and explainable LLM outputs.
• Hands-on experience establishing evaluation frameworks to measure RAG quality and performance.
• Experience building tool-using agents leveraging function calling and API integrations.
• Proven ability to design and implement multi-step agent workflows with safe and controlled execution patterns.

Preferred / Nice-to-Have Skills (Strong Value Add)

• Experience with vector databases and search platforms (e.g., Pinecone, Milvus, Weaviate, Elasticsearch/OpenSearch vector, Azure AI Search, FAISS).
• Familiarity with agent frameworks/orchestration (e.g., LangChain, Semantic Kernel, LlamaIndex) and workflow engines for controlled execution.
• Experience with LLMOps tooling: prompt/version management, evaluation harnesses, observability, A/B testing, red teaming.
• 3+ years of industrial R&D with publications/patents/patent applications.
• 3+ years experience in:
• robotics/automation (including reinforcement learning),
• optimization theory (including black-box optimization),
• designing IoT algorithms under resource/power constraints.
• Cloud experience (Azure/AWS/GCP), containerization (Docker), and scalable deployment patterns (Kubernetes).

Behavioral Competencies

• Strong ownership mindset; proactive in identifying new opportunities and leading initiatives.
• Ability to reconcile competing priorities and deliver pragmatic solutions.
• Collaborative team player with an innovation-first approach.