Synthetic Intelligence is altering industries quick. This shift is powered by giant language fashions, sensible agent methods, and nearer human–AI teamwork. From voice assistants that catch delicate that means to AI groups fixing large duties, we’re transferring from single fashions to related, sensible methods.
In in the present day’s whirlwind of speedy change, it’s simple to really feel lost-unsure the place to be taught, who to comply with, or the best way to really perceive what’s unfolding round you. The true problem? Discovering the correct people- these quietly shaping the longer term in area of interest, cutting-edge fields. And discovering them multi function place? That’s virtually unparalleled.
Till now.
We’re internet hosting a one-stop expertise designed to chop by the noise. An area the place business consultants maintain deep-dive hack periods, providing readability, course, and real-world insights you received’t discover in scattered blogs or countless movies.
You present up. We make it price your time- and your funding. Easy as that.
DataHack Summit 2025 brings collectively world consultants shaping this future. The periods combine deep technical concepts with real-world drawback statements. Whether or not you’re an engineer, AI skilled, product chief, or simply an AI fanatic, these periods will present you the place the sector is heading.
This convention is extra than simply talks and keynotes. It’s a mixture of daring concepts, actual connections, and hands-on studying. With audio system from large giants like Microsoft, NVIDIA, AG2, and extra, you’ll get inside views on AI fashions, protected autonomous methods and the way people and AI can create collectively.
Listed here are ten hack periods that supply each contemporary considering and insightful takeaways you should utilize immediately in the event you select to attend these periods.
AI Voice Agent: The Way forward for Human-Laptop Interplay
Voice is rapidly beating screens as the simplest approach to make use of tech. This session explores how new voice brokers mix generative AI, speech tech, and real-time synthesis to make easy interactions. Audio system share actual use instances, from bots dealing with powerful inquiries to well being assistants serving to sufferers. You’ll learn the way system design impacts the velocity and pure circulate of the brokers. By the top, you’ll know the best way to construct voice apps that really feel extra human.
You’ll additionally see how AI fashions and speech instruments come collectively to energy trendy voice brokers. The session reveals how sensible system design retains solutions quick and correct. You’ll hear tales from buyer help and healthcare, the place voice AI is making a distinction. Lastly, you’ll get ideas for coaching brokers to sound extra actual, serving to you create higher, extra human conversations.
Audio system
- Manoranjan Rajguru – AI Architect, Microsoft
- Ranjani Mani – Director & Nation Head (Generative AI, India & South Asia), Microsoft
Uncover extra particulars right here
A Visible Information to Consideration Mechanism in LLMs
Consideration is what helps giant language fashions concentrate on the correct phrases on the proper time. This session breaks it down with easy diagrams that present how question, key, and worth matrices work. You’ll see visible examples, like gravity models- that assist make these complicated concepts simple to know. The session explains why consideration is essential to coherence and protecting context. It’s nice for anybody who needs to actually perceive considered one of AI’s most helpful instruments.
By the top, you’ll know the way consideration helps generate fluent, related textual content, and the way queries, keys, and values form what the mannequin focuses on. Visible comparisons make the maths simpler to grasp. You’ll construct a psychological mannequin that helps you debug and fine-tune LLMs extra simply.
Speaker
Luis Serrano – Founder and Chief Schooling Officer, Serrano Academy
Uncover extra particulars right here
From LLMs to Agentic AI: Fixing New Issues with Multi-Agent Techniques
Consider LLMs not as lone employees, however as a team-each agent has a activity and ability. This session reveals how multi-agent methods divide duties, enhance flows, and resolve issues that one mannequin can’t. You’ll be taught why single LLMs battle with complicated jobs and the way instruments like CrewAI assist brokers plan, act, and verify outcomes. Actual use instances present how these methods adapt when objectives change. You’ll go away able to construct groups of AI brokers that may develop and work collectively.
You’ll additionally be taught the distinction between single fashions and agentic methods, and the best way to cut up jobs throughout specialised brokers. The session reveals you the best way to use CrewAI to handle these brokers and shares examples that show how scalable and versatile this setup is.
Speaker
Alessandro Romano – Senior Knowledge Scientist. Kuehne+Nagel
Uncover extra particulars right here
Quantifying Our Confidence in Neural Networks and AI
Trusting AI blindly can result in large errors, particularly when the stakes are excessive. This session seems to be on the key query: “how certain is certain sufficient?” You’ll see easy neural nets make daring errors. Then, you’ll be taught two methods to measure how assured fashions are-Bayesian and ensemble strategies. Lastly, the session reveals the best way to use these instruments in bigger fashions like LLMs. You’ll stroll away with instruments to make your AI extra sincere and safer.
The discuss explains why AI confidence issues, and walks by two sturdy methods to measure it. You’ll see the best way to use them on each small and large fashions. It additionally shares concepts for placing these instruments into actual methods so your AI works higher and safer in the true world.
Speaker
Joshua Starmer – Founder and CEO, StatQuest
Uncover extra particulars right here
Agentic AI Meets Accountable AI: Designing Secure Autonomous Techniques
Good brokers can act, assume, and sense however in addition they want guidelines. This session reveals the best way to add equity, openness, and management into agentic AI. You’ll dive into actual points like objectives going incorrect and peculiar behaviors that weren’t deliberate. Case research present the best way to monitor brokers’ lives and implement security guidelines. By the top, you’ll know the best way to construct AI that not solely works but in addition follows values we care about.
You’ll additionally be taught the core moral concepts that information agentic methods and the best way to keep away from widespread issues like system drift or misbehavior. The session offers you instruments to observe brokers in real-time and construct methods that keep truthful and clear.
Speaker
Praveen Kumar GS – Senior Director, Samsung Analysis
Uncover extra particulars right here
Human-In-The-Loop Agentic Techniques
Even sensible brokers need assistance from folks when issues get tough. This session reveals how combining human planning with AI motion can enhance success. You’ll have a look at system designs that blend human considering with agent velocity, utilizing instruments like LangGraph and Pydantic-AI. Actual examples present how human checks catch errors earlier than they develop. You’ll go away with strong plans for constructing these hybrid methods.
The session dives into the best way to combine human choices into AI methods, and provides an outline of instruments like LangGraph and Pydantic-AI. You’ll see how human steerage reduces agent errors, and be taught good methods to construct methods that stability either side.
Speaker
Deepak Sharma – Senior Machine Studying Engineer, DeepMind
Uncover extra particulars right here
AI isn’t right here to exchange creators-it can help them in increasing artistic horizons. This session reveals ReAct-style setups the place folks begin concepts and AI builds first variations. You’ll discover ways to use open instruments like Steady Diffusion, LoRAs, and IPAdapters to make higher content material. The session additionally reveals the best way to add suggestions loops that use scoring to refine your outcomes. Whether or not you’re in design, movie, or adverts, you’ll discover ways to combine your concepts with AI’s energy.
You’ll construct ReAct workflows for teamwork between people and AI, use instruments like Steady Diffusion and LoRAs, and discover ways to rating and enhance outcomes. This session helps you scale your artistic work utilizing sensible AI pipelines.
Speaker
Dhruv Nair – Machine Studying Engineer, Hugging Face
Uncover extra particulars right here
Constructing Actual-Time Multi-Agent AI for Public Journey Techniques
Fashionable cities want transport methods that may react and adapt quick to totally different conditions. This session reveals how multi-agent AI can modify routes and schedules in actual time. You’ll see how Microsoft’s setup lets brokers spot points, provide new routes, and maintain passengers knowledgeable. You’ll additionally be taught design ideas for methods that keep sturdy underneath strain. Should you care about sensible metropolis journey, this session has real-world plans you should utilize.
You’ll discover ways to construct brokers that maintain watch over transit, the best way to react to issues as they occur, and the best way to alert folks immediately. It additionally covers Microsoft instruments that make these methods scalable and dependable.
Speaker
Manpreet Singh – Knowledge & Utilized Scientist II, Microsoft
Uncover extra particulars right here
Agent to Agent Protocol: Advantages and Workflows
For brokers to work collectively, they want a typical language to speak. The Agent-to-Agent (A2A) Protocol units the principles, serving to brokers share objectives, updates, and decisions. This session covers the protocol fundamentals and reveals how brokers group up, resolve issues, and lift flags when wanted. You’ll additionally get recommendations on the best way to use this setup in your methods.
You’ll be taught what the A2A Protocol is, the way it makes agent teamwork simpler, and the best way to use it to maintain your methods easy and future-ready. The discuss consists of examples of workflows and clear steps for easy integration.
Speaker
Avinash Pathak – Senior AI Engineer, NVIDIA
Uncover extra particulars right here
Agentic Data Augmented Technology: The Subsequent Leap After RAG
Uncover how Agentic Data Augmented Technology (KAG) takes RAG to the subsequent degree! Whereas RAG improves fashions by retrieving paperwork, KAG makes use of clever brokers to construct and navigate data graphs for deeper insights. On this session, discover strategies to rework unstructured textual content into structured data graphs and combine them with graph databases for richer mannequin inputs.
See firsthand the constraints of conventional RAG and the way KAG bridges these gaps. The workshop covers end-to-end workflows, together with graph development, database linking, and agent-based reasoning—all powered by LangGraph. Be taught to construct AI methods that don’t simply retrieve info however really perceive context by leveraging dynamic data graphs.
By the top, you’ll know the best way to deploy multi-agent methods that analyze interconnected information sources, unlocking smarter, extra interpretable AI options
Speaker
Arun Prakash Asokan – Affiliate Director- Knowledge Science, Novartis
Uncover extra particulars right here
Conclusion
DataHack Summit 2025 is extra than simply an AI conference- it’s the place the way forward for AI takes form. These periods don’t simply educate; they discover what’s subsequent. With audio system from high corporations and real-world use instances throughout industries, you’ll get a front-row have a look at how AI is fixing complicated issues in the present day and the place it’s heading tomorrow.
On this alternative to hitch such insightful periods in individual, you’ll go away with extra than simply notes. You’ll shift how you concentrate on human–AI collaboration, how fashions can evolve, and the way sensible methods will be each helpful and accountable. This Summit isn’t about watching traits, it’s about being a part of them. In order for you sensible instruments, contemporary concepts, and a peek at what’s coming subsequent, that is the place you’ll discover it.
With a profession spanning 5 years, Swati has labored with high establishments to curate content material for them.
A software program engineer, information science fanatic, and a stickler for grammar who at all times jumps at each likelihood to discover meals, life, and mysterious theories. A Potterhead with a ardour for Psychology, she is at all times up for studying new theories on life. She finds herself amused with books, podcasts, and vlogs that add worth to life.
Login to proceed studying and luxuriate in expert-curated content material.