Senior Software Engineer (AI Focus) • CTO • Technical Architect
With nearly two decades of engineering experience across Big Tech, gaming, enterprise platforms, and AI-first startups, I specialise in designing and delivering scalable backend systems, secure cloud-native architectures, and production-grade AI solutions. My background spans backend system design, microservices development (Python/FastAPI, Java/Spring Boot), API engineering, and distributed systems — complemented by hands-on experience building and orchestrating agentic and multimodal AI applications. From architecting India’s first large-scale e-commerce gamezone platform at Flipkart, to building game platforms at MechMocha and shipping gameplay code at EA Games, and advancing AI accessibility research at King’s College London, I bring a blend of enterprise-grade engineering discipline and rapid innovation. Currently at Areeva.ai, I build AI/ML-driven conversational platforms enabling autonomous AI agents to conduct real-time voice/video interactions — deploying scalable AI workloads on AWS (EKS, ECR, Karpenter) with a focus on productionising LLM systems through CI/CD, monitoring, and model evaluation frameworks.
Dwitee's AI
Digital Twin
Based in London, UK • dwitee.panda@gmail.com
Featured Projects
Real systems, shipped to users
ReMindMap — AI video → summary + mind map
Accessibility tool (BBC R&D, CA11y) that converts video to subtitles, generates simplified summaries, and interactive mind-maps for users with comprehension challenges.
Impact: Prototype validated with test users; auto-narration & live summarization.
HelloPlay / Flipkart GameZone
Kafka + Kubernetes microservices for presence, matchmaking, leaderboards, rewards; integrated into Flipkart app.
Impact: 10M+ downloads, ~3M MAU; improved reliability/latency with observability.
Hello Doctor (Signposting NER for early disease cues)
DistilBERT-based named-entity recognition pipeline to surface early indicators from text for signposting.
Impact: Functional prototype; clean dataset pipeline and demo.
Experience
Companies & roles
Lead Product Engineer (AI Focus) — Contract — Areeva.ai
Jan 2025 – Present- Designed and implemented an Agentic AI platform enabling autonomous AI agents to conduct and evaluate real-time voice/video interview interactions.
- Built self-hosted LiveKit Voice Agent services on AWS EKS — real-time audio STT transcription, LLM response generation, and TTS audio replies at scale.
- Developed multi-agent evaluation pipelines using LLMs to analyse conversations and produce structured scoring, feedback, and candidate ranking metrics.
- Containerised backend services with Docker; deployed scalable microservices on AWS EKS with Helm-based CI/CD via GitHub Actions.
- Designed Kubernetes HPA + Karpenter autoscaling to handle tens-to-hundreds of concurrent AI interview sessions with dynamic cost optimisation.
- Implemented observability stack using telemetry, centralised logging, and runtime monitoring for AI agent reliability.
CTO — Skeptiq
Aug 2025 – Jan 2026- Led end-to-end technical architecture for AI-driven products — GenAI solutions using Python, LangChain/LangGraph, RAG, embeddings pipelines, and agentic workflows.
- Built production-ready LLM pipelines integrating Google ADK for multi-agent orchestration, tool execution, and secure model interactions.
- Designed cloud-native AI infrastructure on GCP (Cloud Run, Vertex AI, Cloud Functions); evaluated AWS with Karpenter for autoscaling and governance.
- Implemented MLOps and CI/CD — automated evaluations, model versioning, observability (Grafana/Prometheus), and standardised LLM deployment workflows.
- Architected microservices for LLM reasoning, agent collaboration, and RAG-based retrieval with fault tolerance and compliance-ready design.
MSc Research Project — Advanced Software Engineering — King’s College London
Sep 2024 – Sep 2025- Built an end-to-end GenAI pipeline combining Whisper (STT), LLM summarisation (T5, Gemini, Mistral/Zephyr), and structured mindmap generation with deterministic JSON/graph output.
- Experimented with multiple transformer models evaluating latency, summarisation quality, coherence, hallucination rates, and prompt optimisation strategies.
- Engineered a production-ready full-stack solution — Next.js frontend + Flask/Python backend deployed on GCP (Cloud Run, Vertex AI); evaluated Azure OpenAI and Azure AI Search for enterprise interoperability.
- Open-source AI/ML research: NER, NLP, RAG on Healthcare; aphasia-friendly mindmap generation using T5, BART, Whisper, and agentic frameworks.
Technical Architect / Software Engineer — Contract — Freelance
Jan 2023 – Jul 2024- Designed and delivered GenAI pipelines using Python, PyTorch, Hugging Face, Vertex AI, and custom LLM orchestration for real-time reasoning, dialogue, and adaptive agent behaviour.
- Integrated transformer-based models (BERT, DistilBERT, Gemini, multimodal LLMs) and diffusion models into production workflows via secure API layers and microservices.
- Evaluated Azure AI ecosystem (Azure OpenAI, Cognitive Services, AI Search, Azure Functions) against GCP pipelines for cloud-agnostic, enterprise-compliant design.
- Built full-stack systems using React.js, Next.js, C#, Golang, Python; cloud-native deployments on GCP Cloud Run with K8s/Karpenter autoscaling and Grafana observability.
- Mentored engineering teams on GenAI-first architecture, structured prompt engineering, and ethical AI design.
Software Development Engineer 3 — GameZone — Flipkart
Sep 2020 – Sep 2022- Built and scaled backend systems for Flipkart GamesZone using Java/Spring Boot and Golang with MongoDB, Redis, HBase, and Apache Kafka for high throughput and fault tolerance.
- Led backend delivery for leaderboards, tournaments, rewards, and game currency — seamlessly integrated into the Flipkart Mobile App.
- Increased DAU/MAU and session duration by onboarding multiple games onto the platform; improved reliability via Grafana, Kibana, and Alertz observability.
- Spearheaded campaign automation and cross-functional technical leadership to align engineering with product objectives.
Lead Product Engineer → Technical Architect — MechMocha Games
Mar 2017 – Sep 2020- Engineered scalable game server architectures for high concurrent loads using Go, Apache Kafka, Postgres, and Java REST APIs; led Android/Kotlin and Unity3D frontend delivery.
- Spearheaded Jetpack Joyride India — dynamic level generator, boss levels, and analytics integration (Admob, Crashlytics, CleverTap).
- Built real-time A/V chat systems using Agora SDK for multiplayer game environments.
- Scaled HelloPlay to 3M MAU over 10M+ downloads; contributed through MechMocha’s Series B raise.
Senior Software Engineer — IKA
Dec 2015 – Mar 2017- Enhanced IKA_RealOne analytical lab software; integrated AR/VR UI and IBM Watson STT/TTS APIs.
- Improved data collection accuracy and reliability of analytical applications through systematic performance enhancements.
- Collaborated with product management on feature prioritisation aligned with regulatory and customer requirements.
Senior Software Engineer — Jumpstart Games
Aug 2014 – Feb 2016- Built iOS/Android educational games with avatar controllers (run/jump/swim), in-app purchases, and MMO room/buddy chat features.
- Developed level manager, NGUI-based UI/UX, and mission/open-world level integration for children’s learning experiences.
Senior Software Engineer — Electronic Arts
Nov 2011 – Jul 2014- Game Play Software Engineer for The Sims 3, Boggle 3D, Monopoly Bingo (Unity3D), and Theme Park.
- Cross-platform C/C++ gameplay systems using OpenGL, DirectX, and Unity3D; GLSL/particles engine VFX.
- Improved proprietary engine reliability and performance across multiple shipped titles.
Software Engineer — Geometric Ltd
Nov 2010 – Nov 2011- CAD/PLM features; PhysX-based cloth/rigid-body physics; 3D visualisations & collision simulation.
- Simulation software built with C/C++, Rendering Engine, and Physics Engine.
Senior Software Engineer — Boeing x Infosys
Jul 2005 – Oct 2010- Enterprise engineering projects for Boeing — OpenGL, C/C++, CAD applications, 2D Math.
Academia & Research
Degrees, publications, teaching
ReMindMap: AI-Assisted Comprehension via Mind-Map Abstraction for Accessibility
Authors: Dwitee Krishna Panda · Timothy Neate
ReMindMap is an AI accessibility pipeline that converts video content into structured mind maps for users with aphasia and comprehension challenges. The system combines automatic speech recognition (Whisper), transformer-based summarisation (T5, Gemini, Mistral/Zephyr), and deterministic JSON graph generation to produce simplified, navigable representations of spoken content. Evaluated across multiple LLMs on latency, summarisation quality, coherence, and hallucination rates.
Hackathons
What I built, learned, and shipped fast
Upcoming
- Encode AI Coding HackathonNov 2025Focus: Prompt-driven 3D worlds
Past
- Google AI HackathonJun 2025Winner
- Encode AI Coding HackathonApr 2025Finalist
- ElevenLab hackathonSep 2025Finalist
- London AI Hack has won the LiveKit side challengeNov 2025Winner
Tech Stack
Breadth with depth
Demo Videos
Let’s build something great
Open to Senior/Staff roles in AI, gaming, or platform engineering. Also exploring CTO/Founding Engineer roles at ambitious startups.
Upcoming Events I will be attending
Where to meet me (talks, hackathons, meetups)