Clicky.foo
A flagship product surface for exploring Clicky - an offline AI assistant direction built around privacy, local vision-language intelligence, and fast user workflows. This is a product, not this portfolio.
offline AI - agentic systems - cinematic web
Building future interfaces like a developer from 2035.
I am Shashank Kumar Singh, a CS undergrad and applied AI builder designing local-first assistants, agent pipelines, and product-grade interfaces with a futuristic developer soul.
AI / VLMoffline assistants, model routing, local inference
SystemsFastAPI, async pipelines, DB-backed workflows
Frontendcinematic UI, motion design, product launches
कोड से भविष्य बना रहा हूँ / कल्पना से निर्माण तक / डिजिटल दुनिया का वास्तुकार / Building futuristic experiences /
live transmission
A portfolio that feels like it is compiling in real time.
The site moves like a story: boot sequence, AI logs, deployment pulses, magnetic cards, and scroll-triggered reveals tuned to stay smooth on real devices.
localprivacy-first AI experiments
2035cinematic developer identity
3Dcursor-reactive hero system
fastmotion tuned for real devices
shashank@future-lab:~/portfolio
SKstack core
Python
FastAPI
Ollama
VLM
RAG
Agents
React
skill constellation
Orbiting between AI systems, polished UI, and local-first engineering.
I like interfaces that feel alive and systems that actually run. The stack leans practical: Python, async workflows, local LLMs, VLMs, FastAPI, Electron, PostgreSQL, Supabase, Git, and web motion.
Agentic workflows
Vision-language models
Model routing
Docker
pgvector
Linux / WSL
GSAP
Three.js
Product UI
product launch bay
Projects presented like premium products, not repo thumbnails.
Clicky.foo is highlighted as a project/product to explore. The links page is separate, so visitors can also jump into the full Shashank link hub when they want the map.
Clicky Windows
Offline Windows AI assistant using local models, screen understanding, speech I/O, and a Python sidecar designed for privacy-sensitive desktop automation.
LUMIS
Cognitive-bias evaluation engine for LLM outputs and automated decision-making, shaped for compliance-minded AI workflows and structured prompt validation.
Stock Inventory Manager
Retail inventory system built for real workflows: ACID-compliant schemas, low-latency search, dashboard data entry, and local-first operational reliability.
GraphMorph
Hackathon GenUI experiment that turns model output into adaptive interface states, exploring what AI-native frontends can feel like when UI is generated in motion.
signal history
Experience with a builder's bias toward shipping.
Oct 2025 - Nov 2025
AI Intern - Vaidsys Technologies
Worked on AI detection pipelines, inference bottlenecks, and internal LLM tooling for automated data analysis workflows.
Aug 2025 - Sep 2025
Python Intern - Navodita Infotech
Automated backend data workflows, refactored legacy modules, and added error handling to reduce repetitive manual processing.
Sep 2025 - Oct 2025
C Programming Intern - Uptricks Services Pvt. Ltd.
Built C utilities with manual memory management, pointer arithmetic, and core data structures under time and space complexity targets.
connect node
Let the next build start from a clean signal.
Reach me by email, explore GitHub, or open the dedicated links page for the full map of my work.