Akshat
Sharma
Building intelligent systems at the intersection of machine learning, web engineering, and research.
About Me
Active Developer
& Builder
I'm an actively working developer with 3+ years of professional experience, currently building impactful products at RecursX Innovations Pvt Ltd. I have delivered multiple successful projects end-to-end, spanning full-stack development and production-grade AI systems.
My work spans LLM applications, RAG systems, and knowledge graphs — I've built and shipped production systems integrating Neo4j, vector databases, and FastAPI backends, taking ideas from concept to deployed reality.
Alongside my professional work, I hold an 8.22 CGPA in CS (AIML specialization) at Shiv Nadar University, a spot on the Dean's List, and a Google Cybersecurity Professional Certificate. I bring both academic depth and real-world execution to every project I take on.
Expertise
Technical Skills
Eight domains, production experience — here's a detailed look at what I build with and how.
Core Languages
My foundation is strongest in Python, JavaScript/TypeScript, and C/C++. I use Python daily for ML pipelines, scripting, and data processing. TypeScript powers all my full-stack projects with strong type safety. My C/C++ background gives me confidence working close to the metal — including embedded systems and performance-critical applications. I also write Solidity for smart contracts and Bash for DevOps automation.
Frontend Development
I build polished, production-grade UIs across React, Next.js, and Angular ecosystems. I'm comfortable with complex state management, SSR/SSG with Next.js, component architecture, and CSS animation. I've built everything from animated portfolio sites to enterprise Angular applications at Fealty Technologies, always focusing on accessibility, performance, and smooth user experience.
Mobile Development
FeaturedHealthcare workflow app — live on App Store & Play Store
I develop cross-platform mobile applications using React Native and Flutter. My most notable app is Mediflow — a healthcare mobile application available on both the iOS App Store and Google Play Store. It streamlines medical workflows and patient management with real-world users across both platforms. I handle the full pipeline: UI design, API integration, push notifications, and release deployment to both stores.
Backend & APIs
Backend is where I'm most productive at scale. I've built production FastAPI services handling vector DB lookups, LLM orchestration, and knowledge graph queries at Fealty Technologies. I've also built Django REST apps with authentication, file handling, and admin dashboards. I design APIs with security, versioning, and performance in mind — always documented with OpenAPI/Swagger specs.
AI / ML & LLMs
AI/ML is my primary research and professional focus. I've built production RAG pipelines, Knowledge-Augmented Generation systems with Neo4j + GNNs, and LLM-based chatbots with multimodal capabilities. At Fealty Technologies, I integrated vector databases (Qdrant, Weaviate) with FastAPI backends for semantic search systems. Academic work spans LSTM forecasting, classifiers, and neuromorphic computing research.
Databases & Storage
I work across relational, document, graph, and vector databases. For AI applications I reach for Qdrant or Weaviate for semantic search. For relationship-heavy data and knowledge graphs, Neo4j with Cypher is my go-to. Everyday apps use PostgreSQL or MongoDB. I use Firebase Realtime Database and Firestore for real-time mobile and web applications.
Blockchain & Web3
I've designed and deployed Ethereum smart contracts using Solidity and Hardhat covering DeFi patterns — escrow, deposits/withdrawals, and reentrancy guards. My CryptoBank project combines Solidity contracts with a FastAPI backend and Next.js frontend into a full decentralised banking prototype. I also built a blockchain-based document protection system using on-chain hashing and permissioned access control.
DevOps & Embedded
I containerise all backend services with Docker and have worked with Kubernetes for orchestration. Proficient in Linux administration and Bash scripting for automation. On the embedded side, I've built IoT prototypes with Arduino, Raspberry Pi, and STM32 — integrating sensors, edge ML inference, and control loops. I also participated in an RC Craft/UAV challenge with custom flight control software.
Career
Work Experience
Founder & Managing Director
SwasthaSarthi Pvt. Ltd.
- Co-founded a healthcare technology venture developing clinical and patient-facing digital products for the Indian market.
- Own product architecture and technology strategy across the founding product portfolio.
- Set the engineering stack, data-protection posture and delivery process from inception.
Founder & Chief Executive Officer
RecursX Innovations Pvt. Ltd.
- Founded a deep-tech venture that translates published but uncommercialised IoT and machine-learning research into deployable commercial systems.
- Leading development of Saksham, a low-cost assistive IoT and TinyML platform for elderly and differently-abled users, now in final-stage hardware prototype testing.
- Launched Samvaad, an AI sales agent for WhatsApp, now serving 500+ Indian SMBs on a tiered subscription.
- Own hardware architecture, sensor selection and bill-of-materials costing alongside embedded firmware, on-device inference and the cloud telemetry backend.
Technical Lead
SageTide Labs LLC
- Led cross-platform engineering, DevOps and delivery management through the launch of Socialyfy, the firm’s first commercial product.
- Owned integration architecture, CI/CD pipelines, environment management and the release process across the stack.
- Introduced application-security review and hardening practices covering authentication flows, dependency auditing and secrets management.
Technical Lead & Project Manager
Fealty Technologies Pvt. Ltd.
- Led Mediflow, a cross-platform healthcare product, from technical architecture through to its first seed round of INR 30 lakh.
- Designed the prescription and medical-records layer using advanced cryptographic constructions for tamper-evident, privacy-preserving clinical data storage.
- Shipped Mediflow to the iOS App Store and Google Play as a production React Native application, owning release engineering and store-compliance review for both platforms.
- Ran sprint planning, backlog prioritisation and delivery tracking, and acted as the technical counterpart in investor and stakeholder conversations.
AI/ML & Generative AI Engineer
Fealty Technologies Pvt. Ltd.
- Built production Retrieval-Augmented Generation pipelines on Qdrant and Weaviate covering chunking strategy, embedding selection, hybrid retrieval, reranking and grounded response synthesis.
- Designed Graph RAG and Knowledge-Augmented Generation systems over Neo4j knowledge graphs, combining Cypher traversal, vector similarity and graph-centrality signals for multi-hop reasoning.
- Architected LangGraph agent workflows with tool use, persistent state, checkpointing and failure recovery for long-running multi-step tasks.
- Trained and fine-tuned small language models, and built evaluation harnesses for retrieval quality and answer faithfulness that made regressions visible before release.
Full-Stack Developer
Fealty Technologies Pvt. Ltd.
- Delivered enterprise applications with Angular, React and Next.js frontends against Django and FastAPI backends.
- Built authentication, role-based access control and audit logging as shared components reused across internal products.
- Established API contract and integration testing conventions subsequently adopted by the wider engineering team.
Backend Engineer
Fealty Technologies Pvt. Ltd.
- Built and maintained REST services with OpenAPI contracts, request validation and structured error handling.
- Implemented relational and document data models across PostgreSQL, MySQL and MongoDB; tuned queries and introduced Redis caching on high-traffic paths.
- Containerised services with Docker and moved deployments onto CI/CD pipelines backed by automated pytest suites.
Conferred by the Director General, Indore Institute of Science and Technology
Recognised for hackathon mentorship. Invited mentor and jury member for college internal hackathons.
Night Mode Activated
You've entered the dark side of the portfolio
Shipped
Flagship Products
Commercial products I have built and taken to market, across two ventures and two client firms.
Saksham
Assistive IoT & TinyML Platform
Low-cost assistive framework for elderly and differently-abled users, running inference on-device so the system stays useful without a reliable network connection. Sensor fusion and embedded firmware on constrained hardware, paired with a cloud layer for telemetry, alerting and caregiver monitoring.
Samvaad
AI WhatsApp Sales Agent (SaaS)
An AI agent running inside a merchant’s WhatsApp Business number: it answers queries against an ingested catalogue, captures orders, and collects payment via in-chat UPI and Razorpay links. Multilingual with voice-note input (Hindi, English, Hinglish, regional), calendar-backed appointment booking, Meta Business Partner integration and DPDP-compliant India-resident storage.
Mediflow
Healthcare Platform
Live on the iOS App Store and Google Play — a React Native client over a FastAPI backend, with cryptographically secured prescription and medical-record storage backed by tamper-evident audit trails. Raised INR 30 lakh in seed funding; led as Technical Lead and Project Manager.
Socialyfy
AI Social Media & Lead Generation
Full-stack AI platform targeting Indian small and medium businesses, launched as the firm’s first commercial product. Led engineering, integration architecture and release management through launch.
FinRecon
Financial Reconciliation Agent
An agent-driven reconciliation service that matches financial records across sources and surfaces the exceptions that need a human decision, replacing manual line-by-line statement checking.
Portfolio
Selected Projects
A collection of production-grade builds, research prototypes, and open-source contributions.
Dwarpal — AP2 Merchant Endpoint for Razorpay
Makes a Razorpay merchant transactable end-to-end by an AI buyer agent. A deterministic policy kernel decides every money action with no model on the path — enforced structurally by a build-failing test that walks the kernel's import closure for any reachable model client. Hash-chained, offline-verifiable evidence packets per transaction; 633/633 adversarial cases blocked across 63 attack techniques with 0 false positives on 176 benign cases.
GraphRAG Social Intelligence System
End-to-end GraphRAG pipeline for social intelligence that converts unstructured text into structured knowledge graphs and enables multi-hop reasoning via LLMs. Includes indexing workflows, entity-relationship extraction, and flexible querying (global/local/chat) over graph memory for context-aware insights.
Real-Time KAG for Node Classification & Link Prediction
Production-style Knowledge-Augmented Graph system integrating GNN-based node classification and link prediction with real-time graph updates. Features hybrid pipelines combining graph embeddings, LLM reasoning, and API-driven workflows for dynamic inference on evolving graph data.
NOAA COOPS Interactive AI Chatbot
Open-source AI chatbot querying oceanographic datasets with natural language. RAG pipeline combining vector search and structured API queries over NOAA sensor data.
Dynamic Database AI Chatbot
AI assistant enabling natural language queries over arbitrary databases. Integrates Qdrant for semantic search and contextual retrieval with secure architecture.
CryptoBank — Web3 Banking System
Decentralised banking prototype with Solidity smart contracts, FastAPI backend, and Next.js frontend. Implements deposit/withdrawal via Ethereum and Hardhat.
NotesMania — College Notes Platform
Centralised notes-sharing platform built with Angular frontend and FastAPI backend. Includes user authentication and a clean, community-focused UI.
TaxiAdda — Real-Time Cab Booking
Next.js frontend with Firebase Realtime DB for live driver locations and bookings. Deployed demo with realtime tracking functionality using Firebase (GMaps API).
Phishing Website Detection
Extracted lexical and domain features; trained a random forest classifier wrapped in a Django demo. Hackathon 1st-place winning prototype.
KNN Credit Card Case Study
Full preprocessing pipeline, feature engineering, and KNN with cross-validation for credit approval prediction. Deployable inference API included.
Get in Touch
Let's Build
Something Great
Whether it's a collaboration, research opportunity, advisory work, or just a conversation about tech — I'm always open. Reach out any time.
Akshat Sharma
Founder & AI Systems Engineer