ANI VERSE / ANIL SUTHAR

Engineering, explained.

Computer Science Engineering student specializing in Cloud Computing at Alliance University, Bengaluru (2023–2027). My project work includes container deployment on AWS ECS/Fargate, CI/CD automation and ML-assisted pipeline reliability. I am seeking Cloud Computing / DevOps internship opportunities.

Public artifacts, owner-reported work and research scope are labeled separately. Editable fields mark missing facts.

Timeline

  1. 2021 · First code — Wrote my first code
  2. 2023 · Code era — Started B.Tech CSE
  3. 2024 · Cloud era — Exploring cloud computing
  4. 2025 · AI era — Exploring artificial intelligence
  5. 2025 · Projects era — Project work begins
  6. 2026 · Projects era — Project work continues

Main projects

AI / LLM engineering

VYOM 4B

A Qwen3.5-4B adaptation for a Hinglish conversational persona.

Problem
Adapt conversational behavior and tone for a Hinglish assistant.
Technical approach
The public model card documents supervised 16-bit LoRA fine-tuning on persona Q&A and provides Transformers inference code.
My contribution
EDIT(owner): confirm your individual contribution and collaborators.
Technologies
Qwen3.5-4B · LoRA · 16-bit · Transformers · Kaggle T4
Architecture & decisions
Adapter-based fine-tuning is documented. EDIT(owner): explain dataset choices, evaluation and your implementation responsibilities.
Implementation status
Published model · prototype
Known limitations
No benchmark, safety evaluation or production-service claim. The model card documents LoRA, not QLoRA.
Engineering focus
Fine-tuning configuration, dataset design and model packaging are the engineering focus; individual ownership awaits confirmation.
Evidence
Public model card checked on 2026-10-22.

Model card

Traffic / computer vision / QUBO research

QuantumFlow-BLR

In-progress research on hybrid traffic-signal optimization for Bengaluru.

Problem
Investigate traffic-signal decisions under changing traffic conditions.
Technical approach
Computer vision and QUBO are the requested research scope. A working end-to-end implementation has not been verified.
My contribution
EDIT(owner): confirm your individual contribution and collaborators.
Technologies
Research: CV / QUBO · Candidate: Qiskit · Candidate: SUMO
Architecture & decisions
EDIT(owner): identify implemented modules, objective/constraints, baseline and simulation setup. Candidate technologies from owner notes are not labeled as completed integrations.
Implementation status
Research in progress
Known limitations
No verified live traffic feed, deployment, optimization gain or completed computer-vision integration.
Engineering focus
Problem formulation, optimization constraints and experimental evaluation are the intended engineering focus.
Evidence
Owner inventory and requested scope only; code/evaluation evidence pending.

EDIT(owner): add a verified public source or project link.

Applied AI / image workflows

Fashion AI Studio

Owner-described Gradio notebook for SDXL-based generation and editing.

Problem
Explore controllable fashion-image generation and editing workflows.
Technical approach
Owner notes describe a Kaggle notebook with generation, image-to-image and mask-editing tabs.
My contribution
EDIT(owner): confirm your individual contribution and collaborators.
Technologies
Gradio · SDXL family · Kaggle notebooks
Architecture & decisions
Notebook plus interactive controls is the described prototype structure. EDIT(owner): confirm checkpoint, mask flow and implemented automation.
Implementation status
Owner-described prototype
Known limitations
Garment analysis and automation are not verified as completed. No quality or latency metrics are claimed.
Engineering focus
Model integration, input/mask handling and reproducible notebook execution are the engineering focus.
Evidence
Owner inventory only; implementation evidence pending.

EDIT(owner): add a verified public source or project link.

DevOps / cloud / ML delivery

AI-Driven CI/CD Pipeline with Failure Prediction

ML-assisted failure prediction integrated with a reproducible Jenkins pipeline.

Problem
Investigate whether build-risk classification can inform deployment decisions.
Technical approach
The supplied resume reports training a classifier on historical pipeline data, adding a Jenkins risk gate and containerizing the workflow with Docker Compose.
My contribution
Resume-reported: classifier training, Jenkins gate integration and Docker Compose workflow.
Technologies
Python · scikit-learn · Jenkins · Docker Compose
Architecture & decisions
Separate prediction from the pipeline gate and reproduce the environment with containers, as described in owner notes. EDIT(owner): confirm features, gate policy and validation split.
Implementation status
Resume-reported prototype
Known limitations
No verified accuracy, false-positive rate, production use or deployment improvement.
Engineering focus
CI integration, classifier boundaries and reproducible infrastructure are the engineering focus.
Evidence
Owner resume supplied on 2026-10-22. Repository URL is recorded in the resume, but public access has not been verified.

EDIT(owner): add a verified public source or project link.

Lab

AI integration / Telegram

Gini 2.0

A public-source assistant using Gemini for claim analysis and chat.

Problem
Make AI-assisted claim analysis accessible through a conversational interface.
Technical approach
The repository describes Gemini integration and a Telegram assistant.
My contribution
Resume-reported: built Gini 2.0 for the Google Gen AI Hackathon, Bengaluru, August 2025.
Technologies
Python · Gemini API · Telegram
Architecture & decisions
EDIT(owner): document source-handling choices and your implemented modules.
Implementation status
Public source · runtime unverified
Known limitations
Model-generated analysis is not proof of factual correctness; no accuracy claim.
Engineering focus
API integration and conversational workflow design.
Evidence
Public repository description checked on 2026-10-22.

Source

Web tooling / prompt workflows

Prompt Forge Studio

A public-source Next.js tool for engineering and managing prompts.

Problem
Organize reusable prompts and their refinement workflow.
Technical approach
The public repository describes Next.js 15 and Gemini integration.
My contribution
EDIT(owner): confirm your individual contribution and collaborators.
Technologies
Next.js 15 · Gemini integration
Architecture & decisions
EDIT(owner): confirm persistence, prompt versions and individual contributions.
Implementation status
Public source · runtime unverified
Known limitations
Feature completeness and live-service availability are not verified.
Engineering focus
Web application structure and reusable workflow design.
Evidence
Public repository description checked on 2026-10-22.

Source

Cloud / containers

AWS ECS / Fargate Application

Container migration from EC2 to ECS/Fargate with VPC routing and logging.

Problem
Understand how a container image becomes a networked application service.
Technical approach
The supplied resume reports EC2-to-Fargate migration, public/private subnets, security groups, ALB routing, reusable task definitions, ECR builds and CloudWatch logging.
My contribution
Resume-reported: migration, VPC/networking design, task definitions and build/deployment configuration.
Technologies
Docker · ECR / ECS · Fargate · VPC / ALB
Architecture & decisions
EDIT(owner): confirm implemented networking, IAM, deployment process and cost controls.
Implementation status
Resume-reported prototype
Known limitations
No verified public deployment, uptime, cost or deployment-time claim.
Engineering focus
Container delivery, networking and observability boundaries.
Evidence
Owner resume supplied on 2026-10-22; public repository existence checked. Individual implementation claims are resume-reported.

Source

Bot engineering / persistence

FURSAT Bot

Owner-described Discord bot with commands, moderation and ranking.

Problem
Support community commands and persistent member state.
Technical approach
Owner notes describe Discord.js v14, prefix commands and MongoDB/Mongoose.
My contribution
EDIT(owner): confirm your individual contribution and collaborators.
Technologies
Discord.js v14 · MongoDB · Mongoose
Architecture & decisions
EDIT(owner): confirm command organization, permissions and persistence design.
Implementation status
Implementation status awaiting confirmation
Known limitations
Code availability, deployed functionality and uptime are not verified.
Engineering focus
Event-driven commands, permissions and data persistence.
Evidence
Owner inventory only; implementation evidence pending.

EDIT(owner): add a verified public source or project link.