class SohamPanda:
"""
AI & Automation Engineer — B.Tech student who builds real,
end-to-end systems rather than academic toy projects.
Speciality: connecting multiple technologies into one solution
that eliminates manual effort entirely.
Backed by frontier models (Google AI Pro + ChatGPT) to ship faster.
"""
def __init__(self):
self.name = "Soham Panda"
self.role = ["AI Engineer", "Automation Engineer", "Systems Builder"]
self.location = "India 🇮🇳"
self.email = "sohampanda95599@gmail.com"
# ── Updated with exact academic context ──────────────────────────
self.status = "B.Tech AI & ML @ SGT University → Building in public"
self.strengths = [
"Workflow automation",
"AI agent development",
"System integration",
"Rapid self-teaching",
"End-to-end product thinking",
"Intent-driven development & Vibe-coding", # ← ships ideas fast
]
# ── Exact modern tooling — what hiring managers scan for ─────────
self.stack = {
"language": ["Python 🐍", "PowerShell", "Bash", "HTML/CSS/JS"],
"ai_agents": ["CrewAI", "Voice AI", "LLM APIs", "Autonomous Agents"],
"backend": ["Flask", "Webhooks", "REST APIs", "Twilio"],
"ml_cv": ["Probabilistic Modeling", "TensorFlow",
"Scikit-learn", "OpenCV", "Pandas"],
"infra": ["GCP", "Cloud Run", "Docker", "Git", "Linux"],
}
self.open_to = [
"Freelance AI / Automation projects",
"Research collaborations",
"Full-time AI Engineering roles",
]
def philosophy(self) -> str:
return "Automate everything. Build once. Let it run forever."
def superpower(self) -> str:
return (
"Taking ambitious ideas → working prototypes with minimal resources, "
"leveraging frontier models to move at maximum velocity."
)|
🤖 AI Agents Autonomous systems that perceive, decide, and act — without human hand-holding. From voice agents to multi-step CrewAI reasoning pipelines. |
⚙️ Automation Pipelines End-to-end workflows that replace hours of manual work. Newsletter digests, form processors, data extractors — set and forget. |
🎙️ Voice AI Real-time conversational AI for calls, receptionists, and customer touchpoints. Built with Twilio, webhooks, and cloud-hosted voice models on GCP Cloud Run. |
🔗 System Integration Connecting APIs, databases, webhooks, and services into one seamless backend. The glue that makes complex systems feel simple. |
A fully autonomous AI receptionist. Near-zero cost. Always-on via GCP Cloud Run.
An end-to-end voice AI system that handles inbound calls autonomously — no human needed. Built as a practical alternative to expensive enterprise call-centre software.
| Feature | Details |
|---|---|
| 📲 Missed call handling | Auto-triggers AI agent on missed calls |
| 🔀 Conditional forwarding | Routes calls based on caller data |
| 🎙️ AI voice interaction | Natural conversation via voice AI services |
| 📋 Caller data collection | Captures name, intent, contact info |
| 🔗 Webhook automation | Real-time event-driven backend triggers |
| ☁️ Cloud deployment | Zero-downtime hosting on GCP Cloud Run |
Stack:
One digest. All sources. Zero manual reading.
A continuously running pipeline that collects newsletters and tech news from multiple sources, deduplicates, summarises with LLM APIs, and delivers a clean digest — automatically.
- 🔄 Multi-source aggregation (RSS, newsletters, APIs)
- 🧠 AI-powered summarisation into concise digests
- 📬 Centralised delivery with minimal maintenance overhead
- ⏰ Scheduled execution — runs without intervention
Stack:
Structured data extraction at scale — from raw docs to clean spreadsheets.
Built a workflow system for parsing and extracting structured information from complex government recruitment forms across multiple organisations, then auto-populating Excel sheets with validated data.
- 🔍 Document analysis & field detection
- 📊 Automated Excel population via Python
- 📐 Workflow standardisation across different form formats
- ✅ Data validation and error-checking layer
Stack:
End-to-end AV pipeline: perception → decision → control.
Explored and implemented a full autonomous driving pipeline covering the core AV problem stack — from sensor perception to vehicle actuation — tested inside a simulation environment.
- 👁️ Perception layer (object detection, lane recognition)
- 🧠 Decision making module (path planning, priority logic)
- 🕹️ Vehicle control (throttle, steering, braking)
- 🧪 Simulation-based testing and validation
Stack:
Superpower proved: ambitious idea → working prototype, under pressure, with minimal resources.
|
Role: Builder · Full-Stack AI Prototype
|
Role: Technical Lead · Shortlisted for Prototype Phase
|
|
Role: Builder · In-Person Delivery Under Pressure
|
Role: Finalist · National Finals, New Delhi
|
|
SGT University · Currently Enrolled Core curriculum: Probabilistic Modeling, Computer Vision, Neural Networks, NLP, Data Structures, Cloud Systems. Applied everything beyond coursework into live, deployed projects. |
|
Self-directed · Open to Freelance & Collaboration Shipped multiple end-to-end systems: voice AI receptionist (PVCA), news digest pipelines, document processors, and autonomous driving simulations. All deployed, not just demonstrated. Leverages Google AI Pro and frontier LLMs to move at maximum execution velocity. |
| 🗂️ Repositories | ⭐ Total Stars | 👥 Followers | 🔥 Contributions | 📅 Member Since |
|---|---|---|---|---|
| Check profile | Check profile | 41+ | 148+ | May 2024 |
AI Engineering ████████████████████ Primary
Automation Eng. ██████████████████░░ Primary
MLOps ████████████████░░░░ Growing
Intelligent Agents ███████████████████░ Primary ← CrewAI + LLM Orchestration
Backend Dev ██████████████░░░░░░ Supporting
Applied AI / R&D ████████████░░░░░░░░ Exploring
"I don't just write code — I build systems that replace manual effort permanently,
leveraging the latest frontier models to create autonomous pipelines.
If it can be automated, it should be."
I'm open to freelance AI/Automation projects, collaborations, and full-time AI Engineering roles.
If you have a repetitive problem, a process that wastes time, or an idea that needs an AI backbone — let's talk.
