ML Engineer

The core responsibilities for the job include the following:

Scale Conversational AI in Production:

• Own and scale conversational AI pipelines that already live in production and push reliability, latency, and quality as traffic grows.
• Extend multi-turn conversations, context management, memory, and agentic workflows to handle massive concurrent scale.
• Improve grounding, consistency, and relevance under real production load, not just in demos.

Build and Scale Voice AI:

• Build and scale our real-time voice AI systems, the platform's newest and fastest-growing surface.
• Work across streaming ASR, TTS, turn-taking, and latency to make voice conversations feel instant and natural.
• Unify chat and voice under a single conversational and memory layer, so context carries across modalities.

LLM and RAG Systems at Scale:

• Build and scale retrieval over astrological knowledge, structured data, and user context across LLMs, embeddings, vector search, RAG, and tool calling.
• Push chunking, retrieval, metadata filtering, and ranking quality as the knowledge base and traffic grow.
• Work with closed- and open-source models, and evaluate when to bring capabilities in-house.

LLM Orchestration and Optimization, at Volume:

• Build and harden model orchestration and routing across LLMs by task, quality, latency, and cost, now running at real production volume.
• Build fallbacks, structured outputs, and model-specific workflows that hold up under scale.
• Own token usage, inference latency, and infra cost as first-class metrics, not afterthoughts.

Personalization and Memory, at Platform Scale:

• Build ML/AI systems that use user profiles, conversation history, preferences, goals, and astrological signals to personalize every interaction.
• Own and extend our personal intelligence and memory systems as the user base scales.
• Improve personalization, recommendations, and ranking so the platform gets sharper with every user, not just every release.

Astrology Intelligence:

• Work with structured astrological data and our astrology computation systems.
• Combine deterministic astrology logic, ML models, retrieval, and LLM reasoning to generate personalized insights.
• Build out Kundli, planetary influence, personality, life-area, timing, and recommendation systems.

Evaluation and Quality, in Production:

• Build evaluation datasets and automated tests for LLM and ML systems, covering both chat and voice.
• Measure quality, retrieval relevance, grounding, hallucination, consistency, and personalization, and catch regressions before users do.
• Close the loop between production feedback and offline evaluation to keep improving at scale.

Production ML Engineering Reliability at Scale:

• Build and own ML/AI services from development through production at platform scale.
• Work with Python, APIs, Docker, AWS, databases, caching, queues, and vector databases.
• Own latency, reliability, and system performance for services handling large, real-time traffic.
• Debug production issues and design for scale before they become incidents.

Requirements:

• 3-5 years of experience in machine learning, applied ML, NLP, or GenAI engineering.
• Strong Python and software engineering fundamentals.
• Hands-on experience with LLMs, RAG, embeddings, vector search, or conversational AI, ideally systems that have run in production, not just in notebooks.
• Experience building and scaling ML/AI systems in production, including real-time or low-latency systems.
• Good understanding of ML fundamentals, model evaluation, and experimentation.
• Experience with APIs/frameworks such as FastAPI/Flask, Docker, and cloud platforms such as AWS.
• Strong grasp of system design, caching, databases, latency, and reliability at scale.
• Ability to independently own a problem from design to implementation evaluation on a production scale.

Good to have:

• LangChain, LangGraph, LlamaIndex, or similar frameworks. Qdrant, Pinecone, FAISS, Chroma, or other vector databases. Redis, PostgreSQL, S3, Kafka/RabbitMQ, or similar infrastructure.
• Hugging Face, open-source LLMs, vLLM, or model serving.
• Speech/voice AI experience: ASR, TTS, streaming audio, or real-time voice pipelines.
• Hindi/Indic language NLP or multilingual AI. Recommendation, ranking, or personalization systems. Fine-tuning, LoRA, PEFT, distillation, or RL-based optimization.
• Experience with LLM observability/evaluation tools such as Langfuse, LangSmith, or Phoenix. Interest or experience in astrology, spirituality, or Indian cultural domains.