Software Developer @ Systems Group · Independent research in computational pathology

Dinakar Pathakota

Backend Systems & Computational Pathology

Building deep-learning systems and large-scale image pipelines for digital pathology.

I work on whole-slide image analysis and the infrastructure that makes it run — GPU-accelerated ML systems, gigapixel image pipelines, and graph neural networks. My long-term interest is building clinically useful, deployable models for digital pathology.

View Research Download CV GitHub

Research

Independent Research — Computational Pathology

Self-directed program building toward doctoral work in medical-imaging ML

2026 — Present

WSI inference platform

Running detection models over gigapixel slides — tumour-cell and mitotic-figure detection — covering slide ingestion, tiling, model execution, and region-level result aggregation.

In progress

Lymph-node metastasis detection (CAMELYON)

A deep-learning study on metastasis detection in whole-slide images from the CAMELYON dataset, targeting a preprint.

Targeting preprint

Vertical WSI/ML for veterinary & toxicologic pathology

A technical and market brief on preclinical, non-regulated pathology — gigapixel imaging demands comparable to the clinical setting, with far lower regulatory friction. Moving from scoping into implementation.

Scoping → build

NTechX

Founder (applied research in AI/ML and security)

›Developed a graph neural network model for automated smart-contract vulnerability auditing; contributed the core detection module for an academic manuscript
›Led the venture end-to-end: research direction, model development, and technical execution

Manuscripts & Research Output

Graph-neural-network–based smart contract vulnerability auditing

Detection module contributed to an academic manuscript

ongoing

Deep-learning detection of lymph-node metastases on whole-slide images (CAMELYON)

Study in progress

ongoing

Experience

Software Developer

Systems Group (Hyderabad, India)

Jun 2026 — Present

›Build scalable internal applications and backend systems; ship production services with FastAPI, PostgreSQL, and Docker
›Delivered ProjectFlow, a self-hosted project tracker (FastAPI + PostgreSQL + Next.js) with OTP email auth, Alembic migrations, GitHub Actions CI, and VM deployment
›Build in-house software for the parent company, Saridena Constructions, alongside customized tools for client requirements

FastAPIPostgreSQLDockerNext.jsGitHub Actions

Engineering Intern — Whole-Slide Imaging Pipelines

Evident Microscopy (formerly Pramana.ai)

Internship

›Engineered DICOM-based ingestion and processing pipelines for gigapixel whole-slide histopathology images, supporting both sparse and fully-tiled acquisition modes
›Built GPU assignment and scheduling scripts to distribute tile-processing workloads across devices, improving throughput on large slide volumes
›Designed RabbitMQ message routing across Python microservices to coordinate acquisition, tiling, and downstream image-processing stages
›Contributed C++ acquisition modules interfacing with microscopy hardware, integrated into the end-to-end imaging pipeline

PythonC++DICOMRabbitMQGPU SchedulingLinux

Selected Projects

All projects

WSI Detection Platform

Gigapixel Pathology Inference

Platform for running AI detection models across whole-slide images — pathology model execution over gigapixel inputs with tiled inference and result aggregation.

PythonPyTorchOpenSlideGPUDICOM

NEXUS OS

Bootable AI Operating Environment

ReAct-style agent orchestrator with skill auto-discovery, a permission model, and a FastAPI server — packaged as a bootable Linux distribution.

FastAPILinuxAgentsRAGVision

Local Document-AI Pipeline

On-Prem OCR + LLM Extraction

Fully local pipeline for structured extraction from scanned documents, with canonical item matching and a comparison dashboard, running on a GPU workstation.

pdfplumberTesseractOllamaPython

ProjectFlow

Self-Hosted Project Tracker

Production project tracker with OTP email auth, Alembic migrations, GitHub Actions CI, and VM deployment — shipped for internal and client use.

FastAPIPostgreSQLNext.jsDocker

Systems I Work With

Whole-Slide Imaging Pipelines

WSI scanners capture pathology images that routinely exceed gigapixel resolution. Nothing about them fits in memory, so the work is in the pipeline: DICOM ingestion, tiling into pyramid levels, and distributing tile workloads across GPUs.

On top of that sits inference — detection models scoring individual tiles, then aggregation back up to slide- and region-level results a pathologist can actually read.

wsi-pipeline.sh

Slide ingestion (DICOM)

Tiling & pyramid build

GPU assignment

Model inference

Region aggregation

Storage & retrieval

Technical Skills

ML & Research

Deep learningGraph neural netsRAGAgent orchestration

Medical Imaging

WSI / gigapixelDICOMImage tilingObject detectionDataset curation

Languages

PythonC++JavaTypeScriptSQL

Systems

DockerFastAPIPostgreSQLRabbitMQLinux / systemdGitHub Actions

Cloud & Compute

AWSGCPAzureGPU computeDistributed inference

Education & Credentials

B.Tech (Honors), Computer Science & Engineering

KL University, Hyderabad · 2022 — 2026

›First Class with Distinction — CGPA 8.86 / 10, 202.5 credits. Graduated April 2026
›Specialization in Cyber Security & Blockchain
›Coursework: machine learning, DSA, computer vision & image processing, distributed systems, cryptography & security

Certifications

AWS Certified Cloud Practitioner (CLF-C02)Microsoft Certified: Azure FundamentalsGoogle Associate Cloud EngineerAutomation Anywhere Certified Advanced RPA ProfessionalGitHub Foundations

Leadership & Community

›Head, Cybersecurity Club — KL University
›Founder, NTechX — AI/ML and cybersecurity venture
›Founder / organizer, 00:00 (Zero Hundred Hours) — youth entrepreneurship community
›Two-time hackathon winner