Daxil Vala

Backend & AI engineer, Boston, MA

Hi, I'm Daxil. I work on backend systems and, more recently, LLM agents and retrieval pipelines. I've been an early engineer at Elocity and Medlaunch, and spent time on product teams at Apollo.io and Open, working on everything from EV charging schedulers to lending services.

These days I'm finishing my M.S. in Computer Science at Northeastern. I'm looking for a full-time role from January 2027, ideally somewhere I can own a service end to end and keep learning from the people around me.

Work

2025Medlaunch ConceptsFounding Software Engineer (Co-op)

Healthcare AI startup. May 2025 – Dec 2025, Clearwater, FL.

  • Helped choose the initial stack and built the first version of the product with a small backend team.
  • Built CLARA, an internal AI agent on AWS Bedrock and LangChain that answers hospital policy questions using RAG over the company's data sources.
  • Automated hospital policy document generation by aggregating structured data across entities.
  • Built AWS Lambda pipelines that parse unstructured policy documents into structured data.
  • Designed a polymorphic MongoDB schema so all certification types share one model.
  • Set up Docker and AWS ECS deployments with CI/CD, and mentored junior engineers as the team grew.

TypeScript · NestJS · MongoDB · AWS Bedrock · LangChain · Lambda · ECS

2023 – 24Apollo.ioSenior Backend Engineer

Deals Intelligence team. May 2023 – Aug 2024, Bengaluru, India.

  • Worked on core features of the Deals product.
  • Led the integration of Deals with Apollo Plays, working across several engineering teams.
  • Ran large data migrations on the production MongoDB cluster using horizontally scaled Sidekiq jobs and rolling batch updates.
  • Built Elasticsearch-backed fuzzy search and filtering to speed up deal creation.
  • Built the data aggregation layer behind the deal analytics dashboard.

Ruby on Rails · Sidekiq · MongoDB · Elasticsearch · Redis

2022 – 23Open Financial TechnologiesSoftware Engineer

Fintech platform for SMB banking and lending. Sep 2022 – May 2023, Bengaluru, India.

  • Helped break a legacy monolith into NestJS and TypeScript microservices.
  • Added New Relic observability to the lending services and fixed the database bottlenecks it surfaced.
  • Built a loan integration service in Go for ICICI Bank, starting as a proof of concept for adopting Go on the team.
  • Normalized an overloaded PostgreSQL JSON column into a relational schema.

TypeScript · NestJS · Go · PostgreSQL · New Relic

2021 – 22ElocityFounding Software Engineer

EV charging management platform. May 2021 – Sep 2022, Bengaluru, India.

  • Redesigned the KPI analytics pipeline around precomputed Elasticsearch aggregations instead of in-memory computation.
  • Built a Python/Flask service that schedules parallel EV charging within grid limits, with real-time charger updates over WebSockets.
  • Built Kafka consumers for charger–CPMS communication, ingesting OCPP telemetry into Elasticsearch for fleet monitoring.
  • Reworked the backend to support multi-tenancy and white-labeling.

Python · Flask · Kafka · Elasticsearch · WebSockets · OCPP

Projects

HackathonRareSenseLLM diagnostic aid for rare diseases

Harvard Rare Disease Hackathon

An LLM-based diagnostic aid that extracts phenotypes from clinical notes, ranks candidate rare-disease diagnoses, and suggests related phenotypes to help clinicians reach a confident diagnosis sooner.

Python · Flask · LLMs · NLP · Prompt engineering

CodeDemo

CourseworkLLaMA Fine-Tuning at ScaleMulti-GPU fine-tuning on an HPC cluster

High Performance Computing · Northeastern

A distributed pipeline for fine-tuning Llama 3.2 1B Instruct on the ORPO-DPO-Mix-40k dataset across multiple GPUs on Northeastern's HPC cluster. We used Hugging Face Accelerate for data loading and cross-GPU gradient sync, with 4-bit quantization and mixed precision to keep memory low, which roughly halved training time going from one GPU to four.

Python · PyTorch · Hugging Face Accelerate · BitsAndBytes · TRL (ORPO) · LoRA

Code

Web3Lido SubgraphFaster staking data on The Graph

The Graph Network

An optimized subgraph for Lido Finance on The Graph, restructuring entities and GraphQL queries so on-chain staking data is faster to index and simpler to query.

TypeScript · GraphQL · The Graph · Node.js

CodeExplorer

AnalysisClickstream InsightsHow shoppers move through a store

Data analysis

Clickstream analysis of an online maternity-clothing store, surfacing browsing patterns, product engagement and navigation drop-offs to inform UX improvements.

Python · Pandas · NumPy · SQL · Data visualization

CodeWalkthrough

Tools I use

Languages
Python, TypeScript / JavaScript, Go, Ruby, Java, SQL
Backend
Node.js / NestJS, Flask, Ruby on Rails, REST APIs, Microservices, WebSockets
Data
PostgreSQL, MongoDB, Elasticsearch, Redis, Kafka, RabbitMQ, Sidekiq
Cloud
AWS (ECS, Lambda, Bedrock, EC2), Docker, Kubernetes, CI/CD, GitHub Actions
AI / ML
LangChain, LangGraph, RAG, MCP, Vector databases, Hugging Face, PyTorch, QLoRA

A bit more

I studied electrical engineering at NIT Karnataka and slowly drifted toward the software side, which is where I've stayed. Most of what I know I learned by owning things in production, alongside teams who were generous with their time and patient with my questions.

At Northeastern I've also served on the Graduate Student Government, first helping plan events and later looking after its website. Outside of all that, I like finding small local restaurants and going for evening runs.

2026
M.S. in Computer Science, Northeastern University · GPA 3.76 / 4.00
2021
B.Tech in Electrical & Electronics Engineering, National Institute of Technology Karnataka
2026
Tech Senator, Graduate Student Government
2025
P&O Senator, Graduate Student Government
Certificates
Graduate Leadership Institute, Northeastern University; Basics of Neo4j: Graph Data Science Library

Say hello

Email is the easiest way to reach me, at daxil.work99@gmail.com. Whether it's about a role, a project, or something you think I'd find interesting, I'd be glad to hear from you. You can also leave a note below.