rana zain@Ranazain

## The Problem Every year over 350,000 Americans suffer out-of-hospital cardiac arrest. Survival depends on the first critical minutes, but during CPR the rescuer's hands are physically locked on the sternum, pressing twice a second. Swiping a phone stops compressions. KeepAlive eliminates hands from the equation: the phone sits on the ground, and your voice becomes the interface. ## The Resuscitation Flow When a bystander shouts *"Help! My dad collapsed, he's not breathing!"*, AssemblyAI Universal-3.5 Pro streaming transcribes the panic in real time. A sub-millisecond local router locks the AHA-BLS protocol, speaks the first command, and starts a Web Audio 110 BPM metronome with visual depth guidance. When the rescuer asks *"Did I break a rib?"*, the answer arrives over the beat without pausing rhythm. Once paramedics arrive, the timers freeze and an EMS handover record (NEMSIS v3.5 shape) is read aloud and generated as a QR code for paramedic tablets. ## Two Brains, One Safety Boundary - **Deterministic Engine:** Controls all medical execution via an AHA-BLS state machine and pre-rendered audio. No generative model can alter CPR instructions. - **Generative Companion:** Answers panic doubts via Groq LPU behind a deterministic safety gate: max 18 words during CPR, zero dosage advice, and a mandatory return-to-beat directive. ## Measured Benchmarks & Stack - **Triage routing:** 0.49 ms | **Groq companion:** 0.38 s | **92 passing tests** - **Unified Voice:** 58 broadcast-mastered ElevenLabs clips keyed by asset ID, with dynamic fallback via `/speak` in the identical voice. - **Stack:** AssemblyAI Universal-3.5 Pro Streaming (16 kHz PCM WebSocket), FastAPI, React/Vite, Groq LPU, ElevenLabs, Web Audio API, Render. *Note: 911 CAD dispatch is simulated (MOCK). EMS handover follows NEMSIS v3.5 schema. Hackathon prototype under MIT license.*

### The Problem: The Multi-Million Dollar Monolith Crisis Every enterprise wants the agility, scalability, and independent deployment of cloud-native microservices. However, manual monolith refactoring is one of the highest-risk endeavors in software engineering. Over 80% of enterprise rewrites exceed their budgets or stall entirely, taking 18 to 36 months and millions in consulting fees. Software architects struggle with tangled code dependencies, implicit database foreign keys, and—most critically—the fear of breaking production, because traditional migrations offer zero mathematical guarantees of behavioral parity. ### The Solution: RepoSplit 2.0 RepoSplit 2.0 is an autonomous, multi-agent engine that decomposes monolithic codebases into runnable, fully verified microservices in under five minutes. Powered by seven specialized agents coordinating via a shared Blackboard architecture, RepoSplit blends deterministic compilers with LLM intelligence (IBM watsonx.ai Granite): 1. **Architect Agent:** Performs whole-repository AST parsing, NetworkX Louvain graph clustering, and Tarjan cycle elimination. 2. **Data Agent:** Isolates schema entities, severs cross-boundary foreign keys, and synthesizes Saga / Outbox patterns. 3. **Contract Agent:** Generates production OpenAPI 3.0 specs, gRPC protos, and typed client SDKs. 4. **Strangler Agent:** Designs Envoy/Kong canary ingress routing for zero-downtime migration. 5. **Scaffold Agent:** Emits runnable FastAPI services, Docker Compose environments, and Kubernetes Helm charts. 6. **Parity Agent:** Automatically synthesizes and replays live traffic against both the monolith and the generated fleet, executing semantic JSON differential assertions to prove zero behavioral divergence. 7. **Governance Agent:** Generates an in-toto v1 and DSSE Ed25519-signed cryptographic Migration Passport locking code provenance and verification proofs. ### Target Users & How They Interact RepoSplit 2.0 is built for Enterprise Chief Architects, Lead Platform Engineers, and DevSecOps Teams. Users interact through: - **Interactive 3D Boundary Studio:** An immersive WebGL digital twin displaying decoupled microservice galaxies and severed RPC conduits in real time. - **Human-in-the-Loop Oversight (EU AI Act Article 14):** Architects can drag, reassign, or tune boundaries with instant live recalculation of coupling metrics before approving code generation. - **Single-Click Fleet Export:** Downloads a ready-to-run microservice zip with CI/CD and Helm manifests. ### Why It Is Unique & Groundbreaking Unlike generic AI code assistants that hallucinate code snippets, RepoSplit 2.0 is deterministic and verifiable: - **Proof, Not Faith:** 100% verified differential parity guarantees the new microservices behave identically to the monolith. - **Auditable Provenance:** Every cut, prompt, and test is cryptographically sealed into an enterprise migration certificate. - **Immediate FinOps Impact:** Delivers an average of 58% cloud cost reduction and an 8.4x ROI over manual consulting.

TableGuard-Lite is a safety-aware closed-loop Vision-Language-Action (VLA) system for bimanual robotic table setting. It combines natural-language instructions, three RGB camera views, and a 12-joint robot state to generate coordinated actions for two SO-101 robot arms using SmolVLA. Instead of executing long open-loop trajectories, the system predicts short action chunks, executes only a small portion, and then replans from fresh observations. A runtime safety supervisor checks joint-rate limits, unexpected contacts, collisions, and scene preservation, stopping unsafe behavior before it can propagate. TableGuard-Lite also introduces recovery-oriented learning: states encountered during failed closed-loop execution are captured and reused for targeted fine-tuning. The project includes a successful full reference manipulation sequence, learned closed-loop control experiments, failure-driven recovery training, three-camera perception, and safety-supervised execution. OpenVINO export, numerical-parity verification, and Intel CPU inference were also successfully demonstrated. TableGuard-Lite shows how Physical AI systems can combine VLA intelligence with safety, recovery, and deployment verification for more reliable real-world robotic manipulation.

ORACLE is a 24/7 institutional-grade autonomous multi-agent options trading terminal designed to eliminate emotional bias and automate complex portfolio management. It orchestrates the full lifecycle from pre-market diagnostics and market intelligence to strategy generation, execution, hedging, risk control, and post-trade analysis. ORACLE deploys 7 quantitative strategy engines: Delta-Neutral Theta Iron Condors for high-IV premium harvesting; Broken-Wing Butterflies for asymmetric risk/reward and volatility-skew capture; Calendar & Diagonal Spreads for time-decay opportunities; Directional Vertical Spreads for volume-POC and momentum breakouts; 0DTE Mean Reversion for intraday SPX/SPY premium capture; Catalyst Straddles for earnings and FOMC events; and a Systematic Wheel Engine for recurring cash flow. Its autonomous pipeline uses specialized agents to evaluate macro conditions, market structure, institutional flow, volatility, and portfolio exposure. Strategy Brain applies Tree-of-Thoughts Monte Carlo simulations to generate and score multi-leg strategies across probabilistic scenarios, while Risk Validator adversarially challenges trades. HITL Governance controls capital allocation and Kelly sizing. Trader Agent executes OCC-compliant multi-leg orders through Alpaca with midpoint slippage control. Portfolio Hedge Agent manages Beta-Weighted Delta, while Risk Bodyguard provides 15-second profit-ratchet protection, hard stops, and 0DTE assignment defenses. Post-Trade Memory performs P&L attribution to improve future decisions. ORACLE also features an AI Copilot powered by LangChain RAG. It explains agent decisions and trade theses, surfaces live Greeks, portfolio risk, market context, positions, and execution status, giving traders transparent access to autonomous reasoning while preserving human oversight. ORACLE transforms options trading from rigid rule-based automation into continuously reasoning, risk-aware autonomous intelligence.

AI AgentGate operates as a digital security firewall and human approval system for autonomous AI agents. As modern AI agents gain the capability to independently interact with external APIs, databases, communication tools, and financial systems, they introduce the risk of making unmonitored or catastrophic mistakes in real-world environments. AI AgentGate solves this problem by placing an inline security gate between the AI agent's code and external execution servers. When an autonomous AI agent decides to perform an action—such as sending a mass promotional email blast or modifying database records—it first submits the proposed action to AI AgentGate. Instead of executing the action immediately, AI AgentGate intercepts and pauses the request. The platform then uses AIML API to evaluate the potential risk and operational impact of the action, automatically generating a safety score from 0 to 100. Low-risk actions are allowed to proceed automatically, while high-risk calls are placed on hold. High-risk proposals are instantly routed to an interactive human approval queue accessible via a sleek dashboard or mobile app supported by Natively / Native Builder. A human manager can review the AI agent's reasoning, inspect the potential risk, and choose to approve or reject the request with a single click. Every request, AI evaluation, and human decision is permanently recorded in an immutable audit ledger powered by Supabase and verified with Bright Data intelligence, giving enterprises complete visibility and control over their AI workforce.

MEIA-LAB (Multi-Agent Earnings Intelligence & Alignment Lab) is an end-to-end AI platform that automates the verification of corporate earnings call claims against official SEC filings in real-time. Every quarter, public companies present their financial performance through earnings calls and slide decks, but analysts must manually cross-reference hundreds of pages of SEC disclosures to separate fact from corporate spin—a process that is slow, error-prone, and costly. MEIA-LAB solves this by deploying a team of specialized AI agents orchestrated through LangChain and LangGraph. The ASR Agent uses Whisper to transcribe earnings call audio with high accuracy. The Vision Agent performs layout-aware OCR on presentation slides using PyMuPDF and multimodal LLMs. The Filing Agent retrieves the latest 10-Q and 8-K filings directly from SEC EDGAR and extracts structured financial data. Finally, the Orchestrator cross-examines all sources, scores each management claim for consistency, flags risk discrepancies, and generates source-cited analyst briefs. The platform features a React-based dashboard that surfaces a real-time Consistency Score, a verified claims panel with direct evidence citations, a risk monitor highlighting guidance gaps, and an Evidence Room linking every finding to exact filing pages and transcript lines. All heavy inference workloads including Whisper transcription and high-throughput vector embedding generation for RAG-based filing search via ChromaDB are accelerated on AMD Developer Cloud GPUs using ROCm, delivering massive speedups over CPU execution. MEIA-LAB cuts institutional-grade earnings research from hours to under a minute, helping hedge funds, equity research desks, and compliance teams invest with certainty.

Sentinel AI is a sophisticated 24/7 autonomous platform designed to unify multi-agent intelligence for sales, finance, and security teams. This innovative system continuously monitors the web, providing crucial insights and automatically alerting the relevant teams when significant events occur. It addresses the common problem of companies spending over $13,500 per month on more than five siloed tools that fail to communicate with each other. Sentinel AI aims to replace these fragmented solutions, offering a more cost-effective alternative at $500 per month while effectively connecting the dots between different data streams. The platform's architecture is built around specialized agents: a GTM (Go-to-Market) Agent, a Finance Agent, and a Security Agent, each leveraging Bright Data for data acquisition and Featherless AI for open-source model inference tailored to their specific tasks. For instance, the GTM agent uses Mistral-7B for entity extraction, the Finance agent employs Phi-3-mini for numerical reasoning, and the Security agent utilizes Llama-3-8B for threat categorization. These agents feed into a powerful Cross-Agent Synthesis Engine, powered by Claude Opus and Cognee for agent memory, which identifies non-obvious cross-domain correlations that would be missed by individual tools. Sentinel AI's capabilities extend to automated workflows triggered by its insights. High-severity GTM signals can be pushed to the CRM with talking points, a churn score above 60 triggers Customer Success alerts and retention playbooks, and critical security events generate incident tickets and customer notifications. platform integrates deeply with partner technologies, utilizing all five Bright Data tools for various scraping and unlocking needs, Featherless AI for efficient open-source model hosting, Cognee for building a robust knowledge graph of signals, and TriggerWare for orchestrating automated workflows. /

Our project is an AI-Powered Knowledge Navigator designed to accelerate hackathons and research projects. It leverages multiple GPT-5-powered agents to understand project goals, context, and constraints. The system performs intelligent searches across GitHub, Kaggle, Arxiv, and StackOverflow, filtering datasets, code snippets, and papers relevant to the project. It auto-generates starter code and reusable components, provides progress tracking with actionable next-step recommendations, and enables team collaboration by summarizing progress and generating reports. This multi-agent system streamlines research, accelerates prototyping, and helps participants focus on building rather than searching

We’ve built a thoughtful digital companion designed to help people escape the endless scroll of short videos not by blocking apps or setting strict limits, but by gently understanding and responding to your habits. Rather than just tracking how much time you spend on your phone, it pays attention to how you’re using it how quickly you scroll, how often you switch videos, the mood of the content you’re watching, and more. When it notices you might be stuck in a loop, it steps in with care. This companion doesn’t lecture or interrupt it starts a friendly, meaningful conversation. It might gently encourage you to take a short break, focus on a small goal, or just breathe and reflect for a moment. It’s not about control. It’s about support helping you feel more in charge of your time and attention, while still enjoying the things you love.

Six Matches. One leaderboard. Build on AMD, earn points and compete for prizes.

🌎 Hybrid Hackathon | 💻 Online Build: 12-17 October 2026 | On-site phase: 17-18 October 2026 To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. The AMD Developer Hackathon: ACT III is a hands-on event for developers, founders, engineers and builders who want to push what’s possible with AI on real infrastructure. 📍 If approved for the on-site experience, you'll also have the opportunity to showcase your project live. Travel and accommodation will not be covered.

🌎 Online Hackathon | 💻 Online Build: Sept 26 - Oct 5, 2026 A virtual, global hackathon. Build from anywhere in the world. Dark Factory is the official hackathon of WeAreDevelopers World Congress North America, presented by BAND, but you don't need to be in San Jose to take part. For one week, you build a software factory in BAND Desktop: a band of coding agents that plans work, implements it and checks its own results. 🎟️ Bonus for sign-ups: you are eligible for a free ticket to WeAreDevelopers World Congress North America in San Jose (Sep 23–25). If you do go, stop by the BAND booth: a surprise is waiting for you.

💻 Online Hackathon | ⏳ Build: 48 hours | 📅 September 25–27, 2026 The official IBM Bob 2.0 hackathon. Build what's next in AI-assisted development, alongside the developers and experts shaping it. ⭐ Get help from experts throughout the event 🤝 Join a team or build solo 💰 Prize pool: $12,000 🧑🏻‍💻 Register before the Kick-Off Stream to start building with Bob