About the Company :
Acronotics is a global consulting, services and a product company specialising in Hyper Automation (IPA & Low Code Process Automation), Data Science & Artificial Intelligence (AI). We design, develop and implement cognitive automation solutions by applying a combination of RPA & AI technologies such as ML, NLP, NLG for clients across industries and around the globe.
The company is headquartered in the UK and has subsidiaries in the US and India. The company's offices are in UK (London), India (Bangalore and Pune) and USA (New Jersey).
About the Role :
We are looking for a skilled AI / ML Engineer to help design and implement GenAI-based systems that interface with real-time enterprise data. You will be responsible for developing, fine-tuning, orchestrating, and integrating LLM-powered capabilities such as retrieval-augmented generation (RAG), function / tool calling, and data-grounded Q&A, within the Azure OpenAI ecosystem.
The ideal candidate brings hands-on experience with LLM orchestration frameworks, prompt engineering, embedding models, and integrating AI systems into production-grade Azure-based platforms.
Core Responsibilities :
LLM System Development :
Design and implement LLM-based pipelines, including :
- Prompt engineering
- Few-shot and zero-shot techniques
- Function / tool calling
- Chain-of-thought and structured output generation
- Work with Azure OpenAI, GPT-4, and embedding models for various use cases
- Build conversational flows, decision trees, and fallback logic for copilots or Generation (RAG) :
Develop and optimize RAG pipelines :
Create embedding pipelines (e.g., using text-embedding-ada-002, Cohere, or Sentence Transformers)Chunk and index content from structured and unstructured sources (PDFs, Office files, HTML, etc.)Store and retrieve embeddings using Azure AI Search, FAISS, or WeaviateEvaluate grounding accuracy and relevance scoringMachine Learning Models :
Build, train, and fine-tune time series forecasting models (e.g., XGBoost, Prophet, ARIMA, or LSTM) for financial KPIs where GenAI requires predictive contextCombine structured model outputs with LLM reasoning (e.g., forecasts + narrative insights)Tool / Function Integration :
Integrate structured data APIs, SQL endpoints, Power BI connectors, and OLAP cube access as tools / functions callable by the LLMDesign input / output schemas for safe and deterministic API usage by the modelSupport plugin-style orchestration (LangChain / Function Calling / Semantic Kernel)Evaluation & Iteration :
Define custom evaluation frameworks using metrics like :
Hallucination rateGrounding precision / recallPrompt latency and token efficiencySet up experiment tracking using tools like MLflow, Weights & Biases, or PromptLayerMaintain few-shot / test prompt sets and continuously refineRequired Skills and Experience :
4 to 6+ years of experience in AI / ML / NLP engineeringDeep familiarity with LLM systems : prompt tuning, orchestration, and fine-tuningHands-on experience with :Azure OpenAI ServiceLangChain, Semantic Kernel, or similar orchestration toolsVector databases (Azure AI Search, FAISS, Pinecone)Embedding model APIs (OpenAI, HuggingFace, Cohere, etc.)Strong understanding of time series modeling and ML forecasting techniques in financial domains (e.g., cost, margin, working capital, price volatility)Strong proficiency in Python, with experience in developing modular, testable code for AI / ML pipelines, API integrations, and backend servicesExperience building and deploying backend components (e.g. FastAPI, Flask) to serve AI models or integrate with retrieval pipelinesFamiliarity with best practices for production-grade AI applications, including logging, monitoring, and containerisation (e.g. Docker)Ability to work across the full stack of an AI system from model development to integration and inference APIsExperience in building chatbots or copilots in enterprise settingsKnowledge of Azure cloud services, esp. Functions, App Services, Blob Storage, and Key VaultFamiliarity with enterprise systems like Power BI, SAP, or OLAP cubes(ref : hirist.tech)