RAKSHA RAMESH

Hopeful by design, stubborn by default.

Vol. 1, No. 1Wednesday, September 17, 2025

PROFESSIONAL DOSSIER

A comprehensive investigation into one developer's career trajectory

View PDF in browser

CAREER CHRONICLES

Data Science Intern

June 2025 - Sept 2025

Crum & Forster • Philadelphia, PA

  • Built an OCR pipeline (93% accuracy, 320+ apps/month) consulting with senior stakeholders, cutting review time by ~10 minutes/file and speeding underwriting by 20%.
  • Decommissioned Pony ORM in 10 production microservices, implementing SQL context managers that eliminated connection issues, boosted performance by 20%, and improved reliability for 50+ business users.
  • Created an automated API validation tool (10 services, 500+ payloads) that reduced testing by ~30 minutes/service, standardized edge-case coverage, and earned team-wide recognition.
PythonAWSLLMsOCRGenAI

Software Development Engineer II

July 2022 - May 2023

Walmart Global Tech • Bangalore, India

  • Maintained 5 critical microservices for Walmart US e-commerce, ensuring 2M+ daily transactions ran with 99.9% availability.
  • Led rate limiting rollout validated with peak-traffic simulations and architect reviews; first to implement, cutting outages by 15% (~$15M) and publishing a blueprint adopted org-wide.
  • Spearheaded integration of OpenTelemetry distributed tracing into service workflows, reducing incident resolution time by 20% and enabling faster root-cause analysis across the team.
JavaScriptMicroservicesOpenTelemetryGrafanaCI/CD

Software Development Intern

Jan 2022 - June 2022

Walmart Global Tech • Bangalore, India

  • Designed and built end-to-end integration of PowerSport Category vehicles into product catalogs, adding 1,200+ products and enabling targeted filtering that improved discoverability by 25%.
  • Built and maintained Grafana and Splunk dashboards for 5 microservices, reducing detection time for service issues by 40% and improving uptime reliability.
  • Orchestrated testing and deployment of 2 production services, aligning QA, product, and engineering teams to ensure seamless launches with minimal downtime.
JavaScriptGrafanaSplunkCI/CD

Marketing Intern

May 2021 - Aug 2021

Ellipses Innovation • Bangalore, India

  • Developed data-driven Instagram marketing strategy by analyzing engagement metrics and audience demographics — increasing follower growth by 35% and generating the company’s first recurring sales (20+ orders/week).
  • Conducted competitor benchmarking and market research to refine content mix, posting frequency, and optimal posting times, improving campaign engagement rates.
AnalyticsSocial MediaMarket Research

ACADEMIC ACHIEVEMENTS

Master of Science in Engineering

Dec 2025

Computer and Information Science • AI Concentration

University of Pennsylvania • Philadelphia, PA • GPA: 3.88/4.0

Bachelor of Technology

May 2022

Computer Science and Engineering

PES University • Bangalore, India

AI/ML & NLP PROJECTS

Where ideas come to life

DreamBook — AI Diary Platform

Built AI diary platform digitizing handwritten, audio, and text entries using Google Cloud Vision, Speech-to-Text, and Stable Diffusion (DreamBooth + LoRA) — producing consistent comic-style illustrations.
PythonStable DiffusionGoogle CloudGradio View report

Segment Based Abnormality Detection in EEG Recordings (IEEE, Aug 2022)

Achieved 90.78% classification accuracy (surpassing SOTA) using Empirical Wavelet Transform, Recursive Feature Elimination, and Linear SVM.
PythonMachine LearningSignal ProcessingIEEE Publication View report

BART-to-Edge Translation

Optimized mBART/M2M100 with LoRA and layer freezing, reducing GPU memory by 50% and training time by 30%; enabled sub-1B parameter real-time Zh↔En translation on edge devices.
PyTorchLoRAEdge ComputingNLP View report

Lexical Simplification

Contextual word substitution pipeline improving text accessibility — reduced reading complexity by ~20% and improved comprehension in testing via POS tagging and frequency-based filtering.
NLTKgensimpandasAccessibility View report

Deep Learning for Text Summarization

Implemented extractive and abstractive pipelines; abstractive models improved semantic coherence while reducing repetition by ~25%.
ROUGEBLEUAbstractiveExtractive View report

DATA SCIENCE & ANALYTICS PROJECTS

Optimizing ANN for Learned Cardinalities

Tuned MSCN-based learned cardinality estimator; selected Adam + MLSE for smoother convergence, lowering median Q-error and standardizing workflows.
PyTorchDepp LearningMLSECardinality Estimation View report

World Development Indicators Analysis

Modeled macro growth with regression on World Bank WDI data; achieved RMSE 2.7% (developed) and 4.3% (developing), producing sector-driven insights.
Regressionmulti-disiplinaryMacroeconomicsRMSE View report

YACS — Yet Another Centralized Scheduler

Built master–worker scheduler with Random, RR, and Least-Loaded policies; added map-reduce dependencies and back-pressure to reduce idle CPU time.
PythonSocketsMultithreadingMap-Reduce View report

SYSTEMS & INFRASTRUCTURE PROJECTS

PennOS — Custom Operating System

UNIX-like OS with FAT filesystem, multi-level feedback queue scheduler, and preemptive user-space threading. Custom shell with built-ins and process control syscalls.
CFAT FSMLFQConcurrency View report

Mini-Compiler for JavaScript

Compiler for a JavaScript subset implementing lexical, syntax, and semantic analysis; generated optimized 3-address code with constant folding and DCE; scoped symbol table and type checking.
CLexingParsingThree-Address Code View report

Elevator Control System

C-based simulation using an ordered doubly linked list for prioritized floor requests; direction-aware scheduling reduced average wait times.
CData Structures View report

CONTACT THE NEWSROOM

rakshar@seas.upenn.edu
raksharamesh0002@gmail.com

TECHNICAL ARSENAL

Programming

PythonTypeScriptNode.jsCGraphQLGitPostmanCI/CDJIRAConfluenceTableau

AI & Machine Learning

PyTorchNumPypandasscikit-learnEWTEMDRFELoRADreamBoothStable DiffusionBLEUBERTScore

Cloud & Infrastructure

AWS (CDK, Textract, Bedrock, BDA, Step Functions)Serverless FrameworkGoogle Cloud VisionSpeech-to-TextKubernetesIstio

Data Eng. & Observability

SQLPony ORMGrafanaSplunkPrometheusOpenTelemetryJaeger

Systems & Compilers

FAT filesystemMulti-level feedback queue scheduler

Product & Collaboration

Agile/ScrumMVP developmentUser researchA/B testing

LEADERSHIP DESK

Head Teaching Assistant / Product Manager

University of PennsylvaniaAug 2024 – Present

  • Directed product strategy for 5 cross-functional student teams (agile planning, feedback loops, stakeholder demos).
  • Coordinated a team of 11 TAs; training, task allocation, daily operations; supported 100+ students & faculty.
  • Managed course logistics: scheduling, grading workflows, resource planning; evaluated 20+ demos for market fit.

Chair — English Literacy Program

U&I Trust (Bangalore, India)May 2022 – Jul 2023

  • Directed program for 20+ underprivileged girls; designed lesson plans for conversational & written fluency.
  • Led volunteer recruitment (+50% staff) and fundraising to fully cover annual costs.

Chair — ACM-W Student Chapter

PES UniversityAug 2020 – Jul 2022

  • Expanded membership 4×; secured sponsorships & VC funding for Women’s Day Ideathon (backed top 3 ideas).
  • Organized 15+ workshops, hackathons, and speaker events promoting diversity in tech.

Event Lead / Organizer

PES UniversityAug 2019 – Jul 2022

  • Executed national technical & cultural fests (10,000+ attendees/year); managed logistics and budgets.
  • Led sponsorships, vendor negotiations, and marketing; increased footfall ~20% YoY.

CLASSIFIED AD

Seeking: Complex problems at intersections of tech and humanity

Offering: Unlimited curiosity, and stubborn optimism

Interests: AI for accessibility • creative AI • large-scale system design