Sarankirthic SP
Sarankirthic SP
Software Engineer | Fullstack, Backend & Embedded Systems
Open to opportunities

Hi, I am Sarankirthic. Building systems that scale — from microcontrollers to the cloud. I engineer real-time backends, data pipelines, and embedded solutions that are fast, resilient, and built to last. Whether it's a distributed event-driven platform or a cloud-native API, I bring precision and purpose to every layer of the system.

Education
B.Tech — Electronics & Communication Engineering
2019 – 2023
Vellore Institute of Technology
↳ Graduated with Honors
Work Experience
Software Development Engineer · Seeedstack
July 2025 - Present
  • Architected the end-to-end technical foundation of Lynk from the ground up, defining the core tech stack, system design, and service boundaries to support rapid feature scaling
  • Engineered a cloud-native backend using Flask (Python) and PostgreSQL, with a React frontend, reducing average API response time to under 50ms and maintaining a 99.9% uptime SLA across all production endpoints
  • Designed and implemented a CI/CD pipeline using GitHub Actions and Docker, cutting deployment time by 60% and enabling daily production releases
  • Built a real-time event-driven architecture using Kafka, decoupling 5+ core services and enabling asynchronous processing of 10,000+ events per minute
  • Provisioned and managed cloud infrastructure on AWS (EC2, RDS, S3, CloudWatch), achieving infrastructure cost optimization of 30% through right-sizing and auto-scaling policies
  • Designed a role-based access control (RBAC) system with JWT authentication, securing all API endpoints across 4 user permission levels
  • Owned end-to-end observability by integrating structured logging, distributed tracing, and alerting, reducing mean time to resolve (MTTR) incidents by 50%
Fullstack Engineer · Cranberry.fit
Oct 2024 — July 2025
  • Built Rasa chatbot with WhatsApp integration handling 50,000+ monthly interactions
  • Designed and implemented RESTful APIs for the core platform
  • Optimised database queries improving system performance by 40%
Fullstack Engineer · Temperstack
May 2024 — Oct 2024
  • Integrated AWS WAF with multi-cloud monitoring systems
  • Developed real-time threat detection and response pipelines
  • Implemented automated security workflows reducing incident response time by 60%
Projects
GatePrepare Flask, React, PostgreSQL, Celery, Kafka, Redis, Docker
Problem Statement:

GATE aspirants rely on passive studying — rereading notes, cramming, and re-watching lectures — with no system to track what they've actually retained. This leads to wasted revision cycles, forgotten concepts right before the exam, and no personalized way to know which topics need review versus which are already mastered.

The Idea:

Build a platform that turns raw study material into a structured, personal knowledge vault, then uses proven memory science to tell the student exactly what to review and when — replacing guesswork with a data-driven revision schedule.

Solution:

An AI-native prep platform where a crawl4ai pipeline and pgvector semantic search auto-build a personalized knowledge vault from study material, a SM-2 spaced repetition engine schedules reviews at the optimal recall window, and Claude-powered Q&A gives instant concept clarity — all backed by an event-driven Kafka and Celery architecture.

SevaAI Flask, PostgreSQL, Celery, Redis, Claude API, Gupshup, Sarvam AI, Razorpay, Docker
Problem Statement:

Indian clinics and small healthcare businesses lose 30-40% of potential bookings to missed calls, since front-desk staff can't answer every call around the clock or across every language their patients speak. This translates directly into lost revenue and no reliable way to scale patient intake without hiring more staff.

The Idea:

Build a live, 24/7 AI receptionist that can answer calls, understand intent, and book appointments in the patient's own language — giving small clinics enterprise-grade front-desk coverage with zero human overhead.

Solution:

A multi-tenant AI receptionist architected on Flask and PostgreSQL with row-level security for tenant isolation, Claude tool-use agents for intent resolution, Sarvam AI for multi-language STT/TTS, Gupshup for WhatsApp delivery, and Razorpay for billing — backed by 132 automated tests and a DPDP-compliant audit trail.

Lynk Flask, React, PostgreSQL, Celery, Redis, Docker
Problem Statement:

Job seekers waste hours scrolling through scattered listings across multiple job sites, manually tailoring resumes and cover letters for each application. This slow, fragmented process means strong candidates miss relevant openings simply because applying doesn't scale with the number of jobs worth applying to.

The Idea:

Turn job hunting into a fast, swipe-based experience — let an AI matching engine surface and rank the most relevant live listings, then auto-generate tailored applications so applying takes one tap instead of twenty minutes.

Solution:

A Tinder-style swipe platform that parses resumes and scrapes/ranks live listings across top job sites via an AI matching engine — a right swipe auto-submits a tailored application while a left swipe discards it, turning passive scrolling into targeted, one-tap applying.

Data Analytics Platform Python, Apache Spark, PostgreSQL, Docker, AWS
Problem Statement:

Organizations generate massive volumes of data from multiple sources, but lack a unified system to ingest, process, and visualize it in real time. This leads to delayed decision-making, fragmented insights, and inefficient manual reporting pipelines that fail to scale with growing data demands.

The Idea:

Build an end-to-end data platform that automates the flow of raw data from ingestion through transformation to real-time visualization, eliminating manual bottlenecks and giving stakeholders live, actionable insights.

Solution:

A fully integrated data pipeline that ingests data from multiple sources, processes and transforms it in real time using event-driven architecture, and delivers interactive dashboards and visualizations — enabling faster, data-driven decisions at scale.

E-Commerce Platform Spring Boot, React, MySQL, Redis, Stripe API
Problem Statement:

Small and mid-sized businesses struggle to establish a reliable online retail presence due to the complexity of building secure, scalable storefronts. Existing solutions are either too costly, heavily restricted, or lack robust payment security — making it difficult to serve customers safely and efficiently.

The Idea:

Develop a full-stack e-commerce platform that handles the complete shopping experience — from product browsing and cart management to checkout — with secure, end-to-end payment integration built in from the ground up.

Solution:

A scalable web application featuring product listings, user authentication, order management, and seamlessly integrated secure payment gateways — giving businesses a reliable, ready-to-deploy online store that ensures safe transactions and a smooth customer experience.

Learning Management System Java, Spring Boot, Thymeleaf, MySQL, Spring Security
Problem Statement:

Educational institutions and organizations struggle to manage course delivery, track learner progress, and control access across different user roles — instructors, students, and admins — in one place. Fragmented tools and manual tracking lead to poor learning outcomes and administrative overhead.

The Idea:

Build a centralized LMS that unifies course management, learner progress tracking, and role-based access control under a single platform — giving every stakeholder a tailored, permission-appropriate experience.

Solution:

A comprehensive web-based platform where admins manage users, instructors create and deliver course content, and students track their own learning progress — all enforced through granular role-based access control, ensuring a structured, secure, and measurable learning environment.

Paisley — News Recognition & Automation Python, TensorFlow, Flask, MongoDB
Problem Statement:

The modern news landscape is overwhelmed with content from countless sources, making it nearly impossible for users to filter relevant information, detect patterns across stories, or receive personalized recommendations without spending hours manually sifting through feeds — leading to information overload and missed insights.

The Idea:

Leverage AI to automatically analyze incoming news content, recognize patterns, topics, and trends across sources, and deliver intelligent, personalized recommendations — turning raw news volume into a curated, meaningful experience for each user.

Solution:

An AI-powered engine that continuously ingests news data, applies pattern recognition to identify recurring themes and emerging trends, and serves tailored content recommendations — automating the discovery process and ensuring users always receive the most relevant, timely news with minimal effort.

Kafka Routing Orchestrator Apache Kafka, Java, Spring Boot, Docker, Kubernetes
Problem Statement:

In large-scale distributed systems, messages flowing between microservices often require complex routing logic based on content, priority, or destination. Hardcoded routing rules create brittle, unmaintainable pipelines that struggle to adapt to changing architectures — leading to message bottlenecks, misrouting, and cascading failures across services.

The Idea:

Build an intelligent orchestration layer on top of Kafka that dynamically evaluates incoming messages and routes them to the correct topics, consumers, or services — decoupling routing logic from individual microservices and centralizing control.

Solution:

A Kafka-based routing orchestrator that applies configurable, rule-driven logic to intelligently direct messages across distributed services in real time — improving pipeline reliability, reducing inter-service coupling, and enabling seamless scaling and reconfiguration without redeploying individual components.

Twitter Clone — Social Media Platform React, Node.js, MongoDB, WebSocket, Redis
Problem Statement:

Building a social networking platform that handles real-time interactions — posts, likes, follows, and live feeds — at scale is a significant engineering challenge. Most prototype social apps fail under concurrent user load, lack real-time responsiveness, or cannot efficiently deliver personalized feeds as the user base grows.

The Idea:

Design and develop a scalable social media platform inspired by Twitter's core mechanics, prioritizing real-time content delivery, user interaction, and a distributed architecture capable of handling high concurrency without performance degradation.

Solution:

A full-stack social networking application supporting user posts, follows, likes, and live feed updates — built on a scalable backend with real-time WebSocket-driven interactions, efficient data modeling for feed generation, and an architecture designed to handle growing user demand reliably.

Skills & Technologies
Certifications
PCEP - Python Certified Entry-Level Programmer
Python Institute·2024
PCAP - Python Certified Associate Programmer
Python Institute·2024
Data Structures and Algorithms - Certification
GUVI·2025
Introduction to CUDA Programming
NVIDIA
AWS Certified Cloud Practitioner (CCP) - CLF-C02
AWS·2026
Java Programming & Spring Framework
GUVI·2025
Claude API Developer Certification
Anthropic·2026
Claude Code Certification
Anthropic·2026
Model Context Protocol (MCP) Certification
Anthropic·2026
Docker Certified Associate
Docker Inc.·2024
n8n - Certified Automation Expert
n8n·2025
Python Certified Developer
Coursera·2022
Data Structure and Algorithm
Coursera·2022