MA · Shipped by Allen · Data Engineering & Analytics

Allen
Febi.

I build reliable data pipelines and analytics — from SQL transformations in Snowflake and dbt to lakehouse workflows with Airflow and PySpark. M.S. Computer Science student at Northeastern University, with applied AI and analytics experience at SharkNinja.

Data
Reliable data, useful analytics
SQL transformations, data quality, and pipelines built for production.
View experience
256M
Feedback Rows · SharkNinja
12+
Survey Formats Standardized
2,722
Records · TransitForge Run
18
TransitForge Tests
Seeking Data Engineer / Data Analyst roles · Jan 2027 co-ops & Dec 2027 full-time
01

Featured Data Work

01 · Data Engineering

Consumer Feedback at Scale

At SharkNinja, built Snowflake/dbt transformations to standardize 12+ survey formats and a multi-pass Cortex thematic-analysis pipeline across 256M consumer-feedback rows. Diagnosed a deduplication defect that silently dropped survey responses, restoring data completeness.

SQLSnowflakedbtSnowflake CortexCDCData Quality

Problem

Make heterogeneous survey responses consistent and complete enough for consumer-intelligence analysis.

Data sources

12+ internal survey formats and a 256M-row consumer-feedback corpus at SharkNinja.

Pipeline architecture

  1. Survey responses
  2. Snowflake + dbt transformations
  3. Incremental loads + CDC
  4. Multi-pass Cortex analysis

Data quality

Parsed, standardized, and validated survey formats with soft-delete handling. Investigated a deduplication defect that silently discarded valid responses.

Results

Restored survey-data completeness and delivered transformations and thematic analysis during the Applied AI & Analytics internship (June–August 2026). Internal data and interfaces are not published.

Stack
  • Snowflake
  • dbt Cloud
Data sources
  • Oracle
  • Palantir
  • Salesforce
02 · Data Engineering

TransitForge

Airflow/PySpark lakehouse ingesting six live MBTA and NWS feeds every five minutes. A verified run processed 2,722 records into 138 explainable station-route resilience scores in 5.6 seconds, with replayable snapshots and data-quality quarantine.

PythonPySparkAirflowIcebergParquetMinIO

Problem

Turn changing transit and weather conditions into explainable Boston station-route resilience scores.

Data sources

Six live MBTA and National Weather Service feeds, captured every five minutes as complete paginated snapshots with checksums.

Pipeline architecture

  1. MBTA + NWS feeds
  2. Airflow snapshot capture
  3. PySpark validation + scoring
  4. Iceberg / Parquet on MinIO

Data quality

Capture-level partitioning supports replayable outputs; data-quality quarantine separates invalid records from scoring. Eighteen automated tests cover the pipeline.

Results

A verified Boston run produced 138 scores from 2,722 records in 5.6 seconds. A separate architecture validation used 3.48M NYC trips at up to 95K rows/second with atomic overwrites and time travel; this is not the Boston feed volume.

TransitForge lakehouse architecture from live feeds through Airflow and PySpark to Iceberg
Pipeline architecture · source repository

What I’d improve

Next: measure score stability over longer capture windows and add freshness alerts and documented quarantine-recovery workflows.

02

Applied AI

Data is my focus. In this era, that means knowing AI too.

Models are only as good as the data and contracts around them, so I build AI systems the way I build pipelines: typed contracts, deterministic checks, and evidence you can audit. Here are two projects where agents and LLMs do the reasoning and the data layer keeps them honest.

01 · Decision simulation · multi-agent

Canary Pact

Simulates the consequences of a company decision, such as consolidating data vendors or cutting roles that support critical workflows, before a person approves it.

“Agents reason. Deterministic code calculates. Humans decide.”

What I built

  • A versioned “company twin”: a graph of departments, workflows, systems, vendors, controls and KPIs with typed dependencies and linked evidence.
  • Deterministic simulations comparing acting now, waiting and doing nothing, reporting savings, downstream costs, constraint violations and blast radius.
  • LLM department agents plus a challenger agent; every claim is checked against the twin and its evidence, and validated claims can rerun the simulation.
  • A decision package that a signed-in approver can approve, reject or send back; viewers cannot approve.

Data angle

PostgreSQL stores versioned twins as JSONB alongside relational orgs and memberships, plus runs, sequence-ordered events and approval decisions. Shared Pydantic contracts generate JSON Schemas and TypeScript types.

PythonFastAPIPydanticPostgreSQLNext.jsTypeScriptLLM agents
Stack
  • Python
  • FastAPI
  • Pydantic
  • Next.js
  • TypeScript
Data stores
  • PostgreSQL
  • SQLite
02 · Agent observability · MLOps

AgentWatch

Explains why an AI agent is slow by separating measured worker queue delay from execution time, and tying each result to traces and resource evidence.

What I built

  • Reproducible clean-versus-contended CPU experiments recording OpenTelemetry spans and worker enqueue/start/finish timestamps.
  • A resource sampler for Linux CPU pressure and cgroup throttling at 250 ms intervals, aligned to agent-step windows.
  • A locally hosted Qwen3 model (Ollama) with a real tool-call loop restricted to an allowlisted, argument-validated function.
  • A seven-page Next.js dashboard: run history, trace waterfalls, agent conversations, resource evidence and baseline comparisons.
  • An Airflow dataset DAG (acquire → build → validate → split → version → push → alert) with a TFDV schema gate and Slack alerts.

Data angle

Datasets are versioned with DVC on a Google Cloud Storage bucket; a fresh clone with no Google login pulls a release and reruns the pipeline. FastAPI ingests runs into PostgreSQL with validated contracts and duplicate-safe imports.

PythonOpenTelemetryFastAPIPostgreSQLAirflowDVCGCSNext.jsDockerOllama
Stack
  • Python
  • OpenTelemetry
  • FastAPI
  • Next.js
  • Docker
Data & models
  • PostgreSQL
  • Airflow
  • DVC
  • Cloud Storage
  • TFDV
  • Ollama · Qwen3
03

Shipped Apps

Products I’ve shipped or am actively building — live sites, public betas, and open-source repos.

Beta

DevBar

macOS menu bar, CLI, and MCP tools for iOS simulators and Android emulators — boot, install, screenshot, GPS routes.

macOS
Live

Narratify

Offline-first ebook reader and listening companion for Android and iOS — Readium, local audiobooks, on-device narration.

iOSAndroid
Coming Soon

DriveLink

Mount MTP devices directly in Finder.

macOS
04

About Me

"Reliable analytics starts with complete data, clear models, and pipelines you can explain."

I'm a Computer Science graduate student at Northeastern University focused on data engineering and analytics. My work spans SQL transformations, data modeling, data quality, and production pipelines, backed by hands-on software engineering experience.

Currently a Graduate Research Assistant on the Food ALERT project, leading application, data, and AWS production infrastructure deployed at foodalertgo.com. At SharkNinja, I built consumer-intelligence data pipelines and analytics automation after earning the internship through SharkNinja JailBreak Edu.

Before Northeastern, I worked as a part-time Software Engineer at Siga Infotech shipping a production POS system for UK restaurant chains, and interned at IIT Guwahati building a P2P solar energy trading platform.

LocationBoston, MA
DegreeM.S. Computer Science, Northeastern
Email[email protected]
Phone(617) 606-8268
GitHubgithub.com/all3n2601
LinkedInin/allen-febi
Siteshippedbyallen.com
Livefoodalertgo.com
05

Technical Skills

Data Engineering / Analytics
SQL / T-SQLSnowflakedbtSnowflake CortexAirflowSpark / PySparkApache IcebergParquetMinIOData ModelingETL / ELTCDCData Quality
Languages
JavaPythonJavaScriptTypeScriptSQLDart
Frontend
React.jsNext.jsFlutterHTML5CSS3TailwindCSS
Backend & APIs
Node.jsExpress.jsFastAPIREST APIsWebSocketJWT Auth
Databases
PostgreSQLMySQLMongoDBFirebaseSQLiteT-SQL
DevOps & Tools
DockerGitAWS ECS / FargateRDS / PostGISS3 / GlueCloudWatchCI/CDVercelLinuxJestpytest
AI / ML
PineconeTF LiteSentence TransformersClaude APIVector Search
06

Work Experience

Jun 2026 — Aug 2026
Applied AI & Analytics Intern
SharkNinja · Needham, MA
  • Built Snowflake/dbt transformations to parse, standardize, and validate 12+ survey formats. Added incremental loads, change-data capture, and soft-delete handling for reliable refreshes.
  • Engineered a multi-pass Snowflake Cortex thematic-analysis pipeline across 256M consumer-feedback rows and diagnosed a deduplication defect that silently dropped survey responses, restoring data completeness.
  • Automated cross-system Salesforce localization across 12 markets and OCR tagging for 10,000+ Cloudinary assets, reducing turnaround 60% and achieving 95% tagging accuracy.
Jan 2026 — Present
Current
Graduate Research Assistant
Northeastern University · FoodALERT Project · Boston, MA
  • Owned Food ALERT’s full stack for a five-stage donation lifecycle, designing 15+ normalized PostgreSQL tables and stored procedures and building FastAPI services and Next.js workflows.
  • Built asynchronous risk scoring on six organization and donation signals, auto-flagging high-risk accounts and cutting manual review 60%.
  • Led the Hetzner-to-AWS migration, deploying FastAPI on ECS Fargate and designing the RDS PostgreSQL/PostGIS, ElastiCache, and S3/Glue data layer with backups and CloudWatch monitoring.
Apr 2024 — Jul 2025
Software Engineer, Part-Time
Siga Infotech · Trivandrum, India
  • Managed multi-tenant data for 4 UK retail chains on a shared SQL Server backend, isolating orders, menus, and payments by tenant key across 1,000+ daily transactions and tenant-level reporting.
  • Built pipelines moving order and payment data from the customer-facing (B2C) system into the backend. Rewrote slow T-SQL procedures and indexes, raising throughput from 150 to 250 orders/hour and cutting processing time 40%.
  • Built offline-first SQLite capture with conflict resolution on reconnect, preventing lost or duplicate orders through 4+ hour network outages while keeping 85% test coverage across 500+ tests.
07

Selected Projects

01
Food ALERT
Web
Food ALERT live site hero connecting surplus food with communities
Live product · foodalertgo.com

Production donation lifecycle system with PostgreSQL state machine, FastAPI microservices, Next.js UI, and 6-signal fraud detection.

Next.jsFastAPIPostgreSQLCI/CD
🏆 Hackathon Winner
02
CollabCore
🏆 Hackathon winner
CollabCore Discover Projects dashboard with match scores and recruiting listings
Product UI · discover & match

Student collaboration platform with vector similarity team matching and a normalized backend model; shipped with Docker and AWS CI/CD.

ReactFastAPIPostgreSQLPineconeDockerAWS
03
Restaurant POS System
99.9% uptime (reported)
SIGA POS product page with phone mockup showing weekly revenue and transactions
Production product · SIGA

Production POS shipped for UK restaurant chains with WebSocket order sync, offline-first caching, and T-SQL performance tuning.

FlutterT-SQLWebSocketSQLiteTesting
04
ResPulse
Offline-first emergency prototype
ResPulse mobile app screens showing SOS beacon, map, mesh relay, and offline guides
Product UI · light & dark modes

Offline-first disaster-response platform with BLE distress relays, resilient field workflows, a FastAPI coordination service, and a live operator command center.

FlutterFastAPIBLEReactPostgreSQLWebSocket
05
Arrival
14 destination countries
Arrival app screens: arrival compass, Ask AI, document vault, and budget
Real mobile app · product screens

Privacy-first arrival companion for international students across 14 destinations, combining an encrypted document vault, deadlines, budgeting, and source-grounded guidance.

FlutterNext.jsFastAPIPostgreSQLpgvectorRedis
08

Academic Background

Graduate
M.S. Computer Science
Northeastern University · Boston, MA
Aug. 2025 — Dec. 2027
GPA 3.66 / 4.0
Database Management Systems · Program Design Paradigms · Object-Oriented Software Design
Undergraduate
B.Tech Computer Science
VIT-AP University · Andhra Pradesh, India
Aug. 2021 — Jul. 2025
GPA 8.93 / 10
Data Structures & Algorithms · Object-Oriented Design · Computer Networks · Operating Systems
09

Honors & Publications

CategoryTitleVenue / Detail
🏆Hackathon1st Place — CollabCore Student Collaboration PlatformNortheastern University
🥈Hackathon2nd Place (out of 72 teams) — NavX 24-hour Hackathon CompetitionOpen Competition
🏆Hackathon3rd Place — SharkNinja JailBreak Edu HiFive (VoC Intelligence)SharkNinja
📄PublicationNatural Disaster Management System Co-authorIEEE ICCCSMD'24
🔬PatentSafety Device for Real-Time Environmental Alert System Co-filed · App. No: 202441052914India

Let's
Make
Something.

Seeking Data Engineer and Data Analyst co-ops in Jan 2027 and full-time roles in Dec 2027. Let’s talk about pipelines, data quality, and analytics. Best reached by email.