Building across web, mobile & AI/ML — from production RAG pipelines to cross-platform apps. Startup collaborator. Open source contributor.
I'm Shashank — a B.Tech IT student at RGIPT, building across the full stack. From pixel-perfect web interfaces to production-grade AI pipelines, I care about the craft at every layer.
Currently interning at Myschord (QuizerAI) and shipping cross-platform products. I believe good engineering is invisible — it just works.
Python
Java
Kotlin
JavaScript
TypeScript
C++
SQLPythonJavaKotlinJavaScriptTypeScriptC++SQLPythonJavaKotlinJavaScriptTypeScriptC++SQL
FastAPI
Next.js
React
Flutter
Node.js
Express.js
Three.js
TailwindFastAPINext.jsReactFlutterNode.jsExpress.jsThree.jsTailwindFastAPINext.jsReactFlutterNode.jsExpress.jsThree.jsTailwind
TensorFlow
NumPy
Pandas
AWS
Docker
Kubernetes
RedisLangChainTensorFlowNumPyPandasAWSDockerKubernetesRedisLangChainTensorFlowNumPyPandasAWSDockerKubernetesRedisLangChain
MongoDB
PostgreSQLMySQL
DynamoDB
Git
GitHub
JenkinsMongoDBPostgreSQLMySQLDynamoDBGitGitHubJenkinsMongoDBPostgreSQLMySQLDynamoDBGitGitHubJenkinsArchitected a multi-tenant RBAC engine supporting Student, Teacher, and Admin roles with backend-authoritative subscription gating (Free/Go/Pro) across 3 privilege tiers.
Designed and shipped a production RAG pipeline using LangChain and AWS Bedrock with vector-indexed knowledge base, orchestrated via multi-instance Lambda, scaling to 5,000+ concurrent users at 40% lower inference latency.
Executed a MySQL → DynamoDB migration improving API response by 60%, and containerized microservices via Docker cutting environment bugs by 90%.
Built role-scoped Recharts analytics dashboards and integrated Google OAuth 2.0 with JWT-based session management and RBAC permission propagation.
Launched a full-stack certification platform with chapter-gated quiz progression and an AWS Lambda-orchestrated PDF ingestion pipeline processing 18 chapters at 95%+ parse accuracy into MongoDB.
Designed an ontology-driven NLP purport classifier with token-optimized semantic chunking, cutting AWS Bedrock token costs by 35%.
Implemented a Redis TTL cache layer reducing redundant LLM inference calls by 70% and delivered Recharts dashboards tracking certification conversion and cohort engagement.
Engineered a cross-platform AI productivity app with Flutter/Kotlin Android client and Next.js web frontend, achieving 60% reduction in task-creation friction.
Integrated Sarvam AI STT/TTS for a voice-to-task pipeline with sub-2s processing latency, applying NLP subtask decomposition cutting manual planning overhead by 50%.
Deployed the official RGIPT IEEE chapter platform with event registration workflows, a secure admin CMS, a Canvas API dynamic ID card generator, and a SendGrid email pipeline serving 980+ members.
Open to internships, freelance work, open source collaboration, and helping startups figure out their engineering challenges.
