--- title: "Manas Dasgupta" url: https://manasdasgupta.com/ description: "Hands-on AI engineering courses: LLMs, RAG, AI agents and LLMOps" author: Manas Dasgupta last_verified: 2026-11-12 --- # Manas Dasgupta > **Bottom line:** Manas Dasgupta is a CTO, AI architect, author and trainer who teaches AI engineering: building LLM applications - Harness Engineering, Agentic Application Development, RAG applications with hybrid search and vector databases, Multi-agent systems with LangChain, Deep Agents, LangGraph, OpenAI Agents SDK, MCP and CrewAI, plus AI evaluations, LLMOps and guardrails. He has taught 10,000+ AI professionals across 60+ countries. ## Start with your role - [Software engineer moving into AI](https://manasdasgupta.com/paths/engineer/): You write code and want to build LLM, RAG and agent applications and move into an AI engineering role. - [Data professional moving into AI](https://manasdasgupta.com/paths/data/): You work with data or machine learning and want to add generative AI, RAG and LLM skills. - [Security professional applying AI](https://manasdasgupta.com/paths/security/): You work in cybersecurity or IT support and want to use AI agents, LLMs and machine learning for defence. - [Business professional using GenAI](https://manasdasgupta.com/paths/business/): You work in HR, finance, customer experience, operations or Agile delivery and want to automate your work with generative AI, without coding. ## Courses - [Gen AI - LLM RAG Two in One - LangChain + LlamaIndex](https://manasdasgupta.com/courses/llm-rag-langchain-llamaindex/) (Udemy): Learn to build retrieval-augmented generation (RAG) applications with both LangChain and LlamaIndex. You work with OpenAI GPT and Google Gemini, store embeddings in ChromaDB and Pinecone, build agents, tools, query pipelines and conversational chatbots across 11 hands-on projects, and deploy a LangChain application behind a FastAPI endpoint. - [Introduction to Retrieval Augmented Generation (RAG)](https://manasdasgupta.com/courses/introduction-to-retrieval-augmented-generation-rag/) (Coursera): A three-hour beginner course on retrieval-augmented generation. It explains what large language models can automate, how language embeddings are stored in vector databases, and how to build RAG applications with the LangChain framework through real-world projects. More than 6,600 learners have enrolled, and Coursera issues a shareable certificate. - [Agentic AI for Cybersecurity](https://manasdasgupta.com/courses/agentic-ai-for-cybersecurity/) (EC-Council): A 4h 55m course on building autonomous AI agents for security operations with LangChain and LangGraph. You build a SOC analyst assistant that parses SIEM logs and writes incident reports, add a threat-intelligence knowledge base, create a reconnaissance agent, orchestrate a multi-agent Purple Team simulation, and add human-in-the-loop governance. - [Selecting the Right LLM with Hugging Face](https://manasdasgupta.com/courses/selecting-the-right-llm-with-hugging-face/) (Coursera): An intermediate, three-hour course on choosing the right large language model from the Hugging Face Hub. You learn to navigate the Hugging Face ecosystem, compare models on computational needs, specialisation and practical constraints, use a model in code, and score candidate models' results to pick the best one for a task. - [Troubleshooting Issues with Generative AI](https://manasdasgupta.com/courses/troubleshooting-issues-with-generative-ai/) (EC-Council): A 3h 41m course that teaches tech support professionals to troubleshoot with large language models such as GPT, Gemini and Claude. It compares model capabilities, builds simple RAG applications with vector databases, generates code and commands for support work, and applies LLMs to Windows and Linux troubleshooting. - [Practical Data Science using Python](https://manasdasgupta.com/courses/practical-data-science-using-python/) (Udemy): A 31-hour, project-based course in data science with Python: statistics, exploratory data analysis, NumPy, Pandas and visualisation, then regression, classification, clustering, PCA, feature engineering and hyperparameter tuning with scikit-learn and statsmodels. It ends with deep neural networks in TensorFlow/Keras and deploying models with FastAPI and Streamlit. - [AI-driven Phishing Detection](https://manasdasgupta.com/courses/ai-driven-phishing-detection/) (EC-Council): An intermediate, three-hour course on detecting phishing with machine learning and generative AI. You study phishing attack types through real-world case studies, train and evaluate ML classifiers with feature engineering, detect fake URLs and websites, and use large language models to simulate phishing attacks and build automated detection chatbots. - [Data Science for Cybersecurity](https://manasdasgupta.com/courses/data-science-for-cybersecurity/) (EC-Council): A four-hour intermediate course on applying data science to cyber threats. You collect and preprocess security data, apply supervised and unsupervised machine learning to threat detection, use neural networks and NLP to spot malicious patterns in problems such as malware and phishing, and evaluate models with ethics and regulation in mind. - [Scrum Master's AI Bootcamp: Generative AI as Agile Co-Pilot](https://manasdasgupta.com/courses/scrum-masters-ai-bootcamp-generative-ai-agile-copilot/) (Udemy): A hands-on bootcamp that shows Scrum Masters and Agile practitioners how to use generative AI across the Scrum cycle: generating backlogs and artefacts, planning sprints, running ceremonies and retros, and building custom GPTs. It covers 15+ tools including ChatGPT, Claude, Gemini, Copilot Studio, Zapier and Jira integrations. No coding needed. - [Business Process Modeling with AI](https://manasdasgupta.com/courses/business-process-modeling-with-ai/) (Coursera): An eight-hour beginner course on AI-driven business process modelling. It walks through the BPM lifecycle, shows how AI improves process design, analysis and automation alongside TQM, Six Sigma and Lean, and applies it to real case studies across industries. Offered by Starweaver on Coursera, with 5,300+ learners enrolled. - [GenAI for Talent Acquisition and Onboarding](https://manasdasgupta.com/courses/genai-for-talent-acquisition-and-onboarding/) (Coursera): A three-hour beginner course for HR and talent acquisition professionals on automating recruitment and onboarding with generative AI. You apply prompt engineering to TA tasks, build AI-powered tools with OpenAI GPT and Google Gemini, and bring best practices and ethical safeguards into your automations. Rated 4.8, with 4,000+ learners enrolled. - [GenAI for Automated Financial Reporting](https://manasdasgupta.com/courses/genai-for-automated-financial-reporting/) (Coursera): A three-hour beginner course on automating financial reporting with generative AI. It covers LLM concepts for finance, prompt engineering for financial contexts, and building reporting tools with OpenAI GPT, Google Gems and Anthropic Projects, including data analysis and anomaly detection. No prior AI or advanced technical skills are needed. - [GenAI for Talent Acquisition: Smarter Candidate Screening](https://manasdasgupta.com/courses/genai-for-talent-acquisition-smarter-candidate-screening/) (Coursera): An intermediate, three-hour course on using generative AI for candidate screening: resume analysis, candidate evaluation and interview question generation. You practise advanced prompting, build custom GPTs and Gemini integrations, and use reflection prompts and ethical evaluation to recognise and reduce bias. No programming is required. - [AI for Customer Experience with Chatbots and Analytics](https://manasdasgupta.com/courses/ai-for-customer-experience-with-chatbots-and-analytics/) (Coursera): A four-hour intermediate course on using AI to improve customer experience. It covers hyper-personalisation, chatbots and predictive analytics, shows how to design AI-driven CX strategies with Perplexity, Claude and Google Gemini, and sets out responsible AI guidelines. No programming is required, and 2,000+ learners have enrolled. - [GenAI for Customer Support](https://manasdasgupta.com/courses/genai-for-customer-support/) (Coursera): A three-hour beginner course on applying generative AI in customer support. You practise prompt engineering for service tasks, build a custom GPT from customer service data, connect a Google Gem to Gmail to automate email support, and use an Anthropic project for visual analysis that detects product defects. No previous AI experience is needed. - [GenAI for Onboarding: Automating Employee Journeys](https://manasdasgupta.com/courses/genai-for-onboarding-automating-employee-journeys/) (Coursera): An intermediate, three-hour course on automating the employee onboarding journey with generative AI. It covers personalisation, document management and feedback collection with GenAI tools, how to design an AI integration strategy for existing onboarding workflows, and the ethical guidelines needed for responsible, unbiased use in HR. ## Corporate training Programs delivered or developed for: ABB, Hector Beverages (Paper Boat), Digit Insurance, TIAA (Teachers Insurance), BCIC, NHRD, EY, NHRD Delhi NCR, UpGrad, TNS Foundation, Smartsheet. Details: https://manasdasgupta.com/corporate-training/ ## Latest writing - [Build, buy or wait: an AI decision framework for technology leaders](https://manasdasgupta.com/writing/build-buy-or-wait/): Build an AI capability only when it is core to your product and you can evaluate it. Buy when it is commodity and a vendor can show evaluation results on your data. Wait when neither is true: model capability and prices change fast, and a six-month delay is often the cheapest option. - [Evaluation-first: why AI products should be tested before they are built](https://manasdasgupta.com/writing/evaluation-first-ai-products/): Write the evaluation set before you build an AI feature. It forces agreement on what good looks like, turns quality debates into numbers, and catches regressions whenever a prompt or model changes. Start with 50 real examples and add every production failure back in. - [One-to-ones that surface problems early](https://manasdasgupta.com/writing/one-to-ones-that-work/): A good one-to-one belongs to the report, not the manager. Let them set the agenda, spend most of the time on what is blocking or bothering them, and keep status updates out. Hold it weekly, never cancel twice in a row, and write down what you agreed. ## Questions people ask ### What is AI product development? AI product development is the practice of building product features whose behaviour depends on a machine-learning model, usually a language model. It differs from ordinary product work in three ways: output is probabilistic, quality must be measured with evaluations, and launch plans need guardrails for wrong answers. ### How do you evaluate an LLM feature before launch? Build an evaluation set of at least 50 real inputs with expected outputs, grade responses with a mix of exact checks, model-based graders and human review, and set a pass threshold before launch. Run the same set on every prompt or model change so regressions are caught before users see them. ### Does Manas Dasgupta deliver corporate AI training? Yes. Manas Dasgupta designs and delivers corporate programs in Generative AI and Agentic AI for technical teams, business functions, HR and leadership. Clients include ABB in Sweden, TIAA in the United States, Digit Insurance, Hector Beverages (Paper Boat), EY and Smartsheet, with industry programs through NHRD and BCIC. Send your team size and goals for a proposal.