Pre-Seed Founder, GIANT  ·  Visakhapatnam, India

Bobbili Somanadh
Vasudevara Sasi Sundar

I am a third-year BTech AIML student and the founder of GIANT. I am building FirmRunner, an AI Agent Suite for accounting firms and professional service SMBs. I ship real product. I build in public. I explain AI the way a founder building with it every day actually sees it.

See FirmRunner Read My Writing
About

Founder building in public.
Creator teaching in depth.

Most people explain AI from the outside. I explain it from inside a product that runs on it every day.

I am a third-year BTech student in AI and Machine Learning at PSCMR College of Engineering and Technology, Vijayawada. While most of my peers study AI in theory, I am building a company with it.

GIANT is my pre-seed company. FirmRunner is its first product: an AI Agent Suite that handles the operational work that kills small accounting firms. Client intake, document collection, billing reminders, deadline tracking, and scheduling. No setup. No API keys. No technical team required.

On the content side, I write and teach AI from first principles across LinkedIn, X, and Substack under my newsletter, The Build Layer. I write for students building careers in AI, engineers who want peer-level depth, and business leaders who need to understand what their teams are actually building.

I have 2 Substack subscribers today. I have shipped 15+ AI systems. I am early, honest about where I am, and moving fast.

Company
GIANT (Global Inertia and Negligence Gap), Pre-Seed
Product
FirmRunner, AI Agent Suite for Professional Service SMBs
Focus Markets
US-First, India-Second
Education
BTech AIML, Year 3 of 4, CGPA 8.2, PSCMR College of Engineering
Location
Visakhapatnam, Andhra Pradesh, India
Platforms
LinkedIn ↗ X ↗ Substack ↗ GitHub ↗
Email
sasisundhar2211@gmail.com ↗
The Product

FirmRunner

Pre-Seed · Active Build

A 5-person accounting firm loses roughly $90,000 a year to work no one bills for. FirmRunner is built to close that gap.

30% of every working hour in accounting is non-billable. That is not a calculation problem. It is an operations problem. FirmRunner ships five pre-configured AI agents that handle the work no one wants to do but every firm pays for. Firms get running in under 10 minutes. No API keys. No technical setup. No IT team required.

Pricing: $199 to $499 per month per firm. 10-question onboarding. Zero setup friction.

01
Client Intake
Captures new client data and routes it to the right workflow without manual entry
02
Document Collection
Follows up on missing documents without a single manual email from the team
03
Billing Follow-Up
AI-generated reminders cut invoice payment time by 45% (Intuit data via GIANT)
04
Deadline Tracking
Monitors filing deadlines and alerts the team before anything slips through
05
Scheduling
Handles meeting coordination and calendar management without back-and-forth
$567B
Global SMB AI market by 2032, up from $194.6B in 2024
Credence Research, 2024
71%
Of SMEs use AI only peripherally. The 29% in core adoption are pulling away.
OECD, 2025
$2B+
EvenUp valuation — legal AI agents, single vertical. The professional services opportunity is validated.
October 2025
Reach Out About FirmRunner
Technical Work

15+ AI systems built.
All open. All shipped.

01
Python LLMs NLP
Procurement Agent System

Manual contract review was slow and inconsistent across large procurement document volumes. Built an LLM-driven pipeline for pricing-anomaly detection and structured field extraction from raw contract text.

Standardized outputs across contract samples and reduced human review effort for recurring document types.

GitHub
02
FastAPI LangChain Tool Use
Multitool LLM Agent

Testing multi-step reasoning and tool-calling workflows for LLM agents requires a controlled environment that most existing frameworks do not provide cleanly. Built an API-integrated agent that chains tools, function calls, and reasoning steps via FastAPI.

Reusable testbed for agent behavior and tool integration. Directly informs FirmRunner's agent architecture.

GitHub
03
scikit-learn NLP Flask
AI Resume Ranking System

Recruiter-style resume screening was manual and inconsistent. Different reviewers scored identical candidates differently based on fatigue, not merit. Built an NLP pipeline using TF-IDF features and ML classifiers exposed via a Flask API.

Increased ranking consistency and enabled plug-and-play integration into external hiring workflows.

GitHub
+ 12
RAG MCP Docker CI/CD
More on GitHub

Computer vision pipelines, RAG systems, NLP classifiers, LangChain agents, FastAPI services, and MCP-integrated tools. All Dockerized with CI/CD via GitHub Actions. End-to-end systems, not Jupyter notebooks.

All Projects
Content and Writing

Teaching AI the way
a founder sees it.

LinkedIn
Founder observations and sourced market analysis

150 to 300 words per post. One sourced data point. One observation only someone building in this space would make. No hashtags. No filler.

Weekdays, 8 to 10 AM IST
Follow ↗
X / Twitter
10-tweet threads built from first principles

Every thread starts with a counterintuitive fact. Every tweet is self-contained. The hook tweet works standalone. Three threads a week, plus standalones.

3 threads per week
Follow ↗
Substack
The Build Layer — weekly newsletter on AI and the build

800 to 1,500 words every Wednesday. Prose only. One central idea, explained from first principles and grounded in what I am actually building with FirmRunner.

Every Wednesday
Subscribe ↗
Topics I cover
AI Fundamentals Large Language Models Agentic AI Prompt Engineering ML from First Principles SMB AI Adoption Founder Build Updates AI Industry Analysis RAG Systems Deep Learning Vibe Coding
Experience

Where I have built
and what I have shipped.

2024 – Present
Founder, GIANT / FirmRunner
GIANT (Pre-Seed Stage)

Building FirmRunner from zero. Product decisions, market research, customer conversations, technical architecture, and content strategy. Every function of the company runs through me.

2023 – Present
Independent AI/ML Developer
Self-directed, 15+ systems shipped

End-to-end Python pipelines integrating LLMs, RAG, agents, and MCP. Dockerized environments. CI/CD with GitHub Actions. Led a 3-member GenAI prototyping group covering task assignments, experimentation, and evaluation.

May 2025 – Jul 2025
AI/ML Intern
BITS Pilani, Hyderabad Campus

Designed and implemented practical machine learning solutions. Built and optimized end-to-end models for real-world decision-making tasks under academic supervision at one of India's top technical institutions.

Feb 2025 – Mar 2025
AI Intern, TechSaksham
Microsoft and SAP via Edunet Foundation / AICTE

Selected for a competitive AI internship backed by Microsoft and SAP. Applied AI and software engineering to real-world industry challenges under the AICTE-approved program.

Jan 2025 – May 2025
Undergraduate Research Contributor
PSCMR College of Engineering and Technology

Contributed to an applied AI/ML research paper. Participated in ideation, technical writing, and literature analysis on AI techniques.

Sep 2023 – Present
BTech, Artificial Intelligence and Machine Learning
PSCMR College of Engineering and Technology, Vijayawada

CGPA 8.2. Coursework in LLMs, agentic workflows, deep learning, and full-stack development. Running GIANT in parallel with the degree.

Technical Skills

The full stack
of what I build with.

AI / ML
  • LLMs and prompt engineering
  • Agentic pipelines, tool use
  • RAG and vector search
  • LangChain, MCP integration
  • PyTorch, TensorFlow
  • NLP, embeddings, tokenization
  • Computer vision, OpenCV
Backend and APIs
  • Python end-to-end
  • FastAPI, Flask
  • Node.js, Express.js
  • REST API design
  • OpenAI and Anthropic APIs
  • Function calling workflows
Cloud and DevOps
  • AWS, Google Cloud
  • Docker, containerization
  • Kubernetes fundamentals
  • GitHub Actions, CI/CD
  • PostgreSQL, MongoDB, Firebase
  • MLOps and versioning
Frontend and Data
  • React.js dashboards
  • TypeScript, JavaScript
  • NumPy, Pandas
  • Data visualization, Tableau
  • MySQL, SQL queries
  • Git and GitHub
Certifications and Achievements

Credentials that
earned their place.

JPMorgan Quantitative Research Simulation
JPMorgan Chase
TechSaksham AI Internship
Microsoft and SAP via Edunet Foundation / AICTE
Data Analyst Job Simulation
Deloitte — Tableau, Excel classification
Hackathon Participant
Devnovate and GDG Build With AI
AI Fluency Certification
Verified AI concepts proficiency
Google Cloud Training
Google Cloud Platform — compute, deployment, orchestration
Get in Touch

If you are building
something serious,
I want to talk.

I am open to conversations with investors, early customers in accounting and professional services, fellow founders, and media. I also publish on AI and building in public — reach out if you want to collaborate on content or be featured in The Build Layer.

Send an Email LinkedIn X / Twitter Substack GitHub