Sarthak Pal
Software Developer
4+ years exp
4+ years ML exp
C++
JavaScript
TypeScript
Go
PHP
Java
C
Python
SQL
Node JS
Scala
Data Pipeline & Feature Engineering: 5/10
Experimentation & Evaluation: 4/10
Active 14 days ago
Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
A backend-focused developer at a middle level who produces working data-integration tooling and algorithmic solutions in Python. The strongest proven skill is practical graph-data engineering, shown by the DataLoader's CSV-to-Neo4j pipeline with CREATE CONSTRAINT, batching (IN TRANSACTIONS OF 10000 ROWS) and Graph Data Science PageRank usage in interface.py. There is limited evidence of production-grade security, structured observability, or large-scale system design in the public code.
Technical skills
C++
JavaScript
TypeScript
Go
PHP
Java
C
Python• Middle • 4y+
SQL• Junior • 4y+
Node JS• Junior • 4y+
Scala• Junior
C++
TensorFlow C++
PHP
Laravel• 4y+
Python
FastAPI
Requests
pySpark• 4y+
Node JS
Express
Databases
MySQL
PostgreSQL
GraphDB
Databases
Redis• 4y+
Apache Kafka
Neo4j
AI/ML
LangChain
LlamaIndex
NLP
RAG
ResNet
Pandas
Jupyter Notebook
LLM• 4y+
Sentiment Analysis• 4y+
spaCy• 4y+
TensorFlow• 4y+
Spark
Frontend
Angular
GraphQL
Next.js
tRPC
React.js• 4y+
DevOps
Amazon EKS
AWS
Azure
CI/CD
Datadog
Docker
GitHub Actions
gRPC
Prometheus
Terraform
Vector
WebSockets
Rest API
Vercel
Amazon EC2
Kubernetes
QA
Locust
Pytest
Mobile
Offline-First
React Native• 4y+
Cybersecurity
OWASP Top 10
Timeline
Arizona State University
Master's Degree •
Computer Science
Software Developer Intern
•
Junior
CitiGroup
•
Internship
Built a real-time streaming pipeline using Python producers deployed on Kubernetes to ingest transactional event logs into Apache Kafka, and modeled relationships in a Neo4j graph to support fraud pattern analysis. Designed a two-tier facial verification system on AWS using SQS to buffer requests between an EC2 web tier and an Inception-ResNet-V1 inference backend. Developed a Scala-based Apache Spark analytics application to run spatiotemporal UDFs and hotspot detection over a very large dataset.
Python
Kubernetes
Amazon EC2
Apache Kafka
Neo4j
Spark
Scala
Software Developer
•
Middle
Beganto Software Systems
•
Full-Time
Integrated GPT-3.5 Turbo into a training SaaS platform to generate personalized learning content and improve delivery efficiency. Implemented a secure document management system with Laravel and SQL for client uploads, storage, and AI-driven training customization. Delivered full-stack improvements with React, created a prompt dashboard to manage and test prompts, and built REST APIs with Node.js/Express plus Redis caching to reduce response times on high-traffic endpoints. Developed a React Native mobile app with offline-first synchronization and high-performance UI animations.
Laravel
SQL
React.js
Node JS
Redis
React Native
LLM
Data Science Intern
•
Junior
Paytm
•
Internship
Analyzed large customer and transaction datasets using PySpark, Python, and SQL to identify behavioral trends and actionable opportunities for product and operational improvements. Produced insights for business decision-making and supported ETL optimization with data engineers and analysts. Built and deployed an NLP sentiment analysis pipeline using spaCy and TensorFlow to automate customer feedback categorization and improve ticket resolution efficiency.
pySpark
Pythonsince 2022
SQLsince 2022
spaCy
TensorFlow
Sentiment Analysis
Middle AI/ML Engineer
Confidence: Medium Data-centric
A self-sufficient data-engineering practitioner at a middle level who builds end-to-end ETL and graph-analytics tooling. The strongest proven skill is data pipeline and graph ingestion work as implemented in DataLoader.load_transform_file which reads parquet, performs cleaning/filtering, exports CSV and loads into Neo4j with constraints and transactional batching. There is no evidence of ML model training, model architecture design, experiment tracking, or production deployment automation in the public code.
Model Architecture & Training
How well models are designed and trained
Not evidenced in public code
Data Pipeline & Feature Engineering
5/10
How data is prepared for models
Real ETL and data-preparation work is present: parquet reading, Pandas cleaning/filtering, CSV export, and structured import into Neo4j with constraints and transactional batching.
Experimentation & Evaluation
4/10
How results are measured and tested
A grader-style evaluation harness validates algorithm outputs with multiple test cases and deterministic checks for PageRank and BFS correctness, supporting reproducible verification of results.
MLOps & Deployment
2/10
How models are shipped to production
Basic operational engineering is shown - connection retry loops, Neo4j driver session management and safe graph project/drop - but no deployment, monitoring, CI/CD or model lifecycle tooling.
Computational Efficiency
2/10
How efficiently computing resources are used
Some efficiency-minded choices exist such as early filtering before export and Cypher batching for imports, but there is no profiling, GPU optimization, quantization or advanced efficiency work.
Research Depth & Innovation
Depth of research and new ideas
Not evidenced in public code
Industries
Data & Analytics• Middle
Transportation & Logistics• Middle
Technologies
Databases
Python• Middle • 4y+
SQL• Junior • 4y+
C++
Scala• Junior
MySQL
PostgreSQL
Redis• 4y+
Rest API
gRPC
LangChain
Terraform
Neo4j
GraphDB
Spark
LlamaIndex
Jupyter Notebook
GitHub Actions
Datadog
FastAPI
Prometheus
WebSockets
NLP
Azure
spaCy• 4y+
CI/CD
TensorFlow• 4y+
Pandas
AWS
Docker
Kubernetes
Apache Kafka
LLM• 4y+
RAG
ResNet
pySpark• 4y+
TensorFlow C++
Amazon EKS
Amazon EC2
Vector
Sentiment Analysis• 4y+
Recommendations
- Develop ETL and analytics pipelines for graph-based datasets and implement Neo4j-backed analytics dashboards or reporting services.
- Build data-quality and validation tooling around the existing grader harness to enable CI tests for data pipelines and graph outputs.
- Work on production hardening of graph ingestion - add logging, monitoring, idempotent loads, and a Docker/compose or Kubernetes deployment for the ingestion pipeline.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Middle Backend Developer
Confidence: Medium Data Platform
A backend-focused developer at a middle level who produces working data-integration tooling and algorithmic solutions in Python. The strongest proven skill is practical graph-data engineering, shown by the DataLoader's CSV-to-Neo4j pipeline with CREATE CONSTRAINT, batching (IN TRANSACTIONS OF 10000 ROWS) and Graph Data Science PageRank usage in interface.py. There is limited evidence of production-grade security, structured observability, or large-scale system design in the public code.
API Design
2/10
How well APIs are designed
Minimal API surface - a small Interface class exposes bfs and pagerank but there is no versioning, idempotency, consistent error contract or HTTP API boundaries.
Evidence
project1/interface.py: bfs(start_node, last_node) method
project1/interface.py: pagerank(max_iterations, weight_property) method
Data Layer & Database
5/10
Working with databases
Concrete database work with Neo4j - schema constraint, batched CSV import, and GDS usage show practical data-layer design though there is no multi-step migration history or advanced transaction/isolation tuning.
Evidence
project1/data_loader.py: CREATE CONSTRAINT location_name_unique IF NOT EXISTS
project1/data_loader.py: LOAD CSV ... IN TRANSACTIONS OF 10000 ROWS
project1/interface.py: use of gds.graph.project and gds.pageRank.stream
Scalability & Performance
4/10
Handling load and speed
Some scaling choices - batching imports and filtering data before load plus use of GDS for analytics; no evidence of caching layers, queue-based decoupling or measured performance testing.
Evidence
project1/data_loader.py: trips filtering and pre-filtering before writing CSV
project1/data_loader.py: LOAD CSV ... IN TRANSACTIONS OF 10000 ROWS
project1/interface.py: explicit drop/project of GDS graph around PageRank
System Architecture
3/10
Overall system structure
Basic module separation (data loader, interface, tester) and a grader/test harness - reasonable for a small project but no evidence of multi-service decomposition, config/secret management, or production-grade deployment patterns.
Evidence
project1/data_loader.py / project1/interface.py / project1/tester.py: clear separation of responsibilities
project1/README.md: describes Dockerfile and project parts (non-scoring README used for context)
Security & Auth
2/10
Protecting data and access
Limited security posture - some database constraints but also hardcoded credentials and unencrypted Neo4j driver usage; input validation and token/lifecycle handling are not present.
Evidence
project1/data_loader.py: GraphDatabase.driver(..., encrypted=False) and credentials used in main
project1/tester.py: TesterConnect uses hardcoded credentials in calls
Reliability & Observability
3/10
Stability and monitoring
Basic reliability patterns - connection verification, retry loops and a context manager to suppress noisy logs; lacks structured tracing, metrics, circuit breakers or sophisticated retry/backoff strategies.
Evidence
project1/data_loader.py: main() retry loop with sleep and exception handling
project1/tester.py: suppress_neo4j_notifications contextmanager and verify_connectivity usage
Expertise
Python• Middle
Databases & Vector Storage• Middle
Industries
Transportation & Logistics• Middle
Technologies
Go
Java
PHP
Laravel• 4y+
Requests
Recommendations
- Develop ETL and graph-processing pipelines that load, validate and batch large datasets into Neo4j (data ingestion and GDS jobs).
- Build data-engineering utilities and test harnesses for graph analytics - automated graders, reproducible PageRank/BFS jobs and integration tests against a Neo4j instance.
- Work on backend Python services that expose graph operations with improved security (secret management, encrypted connections) and robust observability (structured logs, metrics, tracing).
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Junior Frontend Developer
Confidence: Medium UI Engineer
A junior frontend-focused UI engineer who produces polished, responsive user interfaces with attention to design tokens and typographic rhythm. The strongest proven skill is CSS-driven UI and visual polish, demonstrated in personal_website/src/app/globals.css which contains design tokens, responsive grid rules and animations. There is little public evidence of complex state management, backend integration, automated testing, or measured performance engineering.
UI Component Architecture
2/10
How interface parts are built
Basic UI styling and global tokens present but no evidence of deliberate component architecture, composition abstractions, or a custom component library.
Responsive & Cross-browser
4/10
Works on all screens and browsers
Solid responsive CSS using clamp(), media queries and grid; shows attention to fluid typography and breakpoints but no advanced cross-browser feature-detection or RTL work.
Performance Optimization
1/10
Speed of the interface
Minimal evidence of performance engineering - only basic Next.js config flags; no bundle analysis, code-splitting strategy, virtualization or measured optimization artifacts.
Accessibility & Semantics
3/10
Usable for everyone
Reasonable semantic HTML and basic accessibility signals (lang, alt, headings), but no ARIA on custom widgets, focus management, or automated a11y checks shown.
State Management & Data Flow
Managing data in the app
Not evidenced in public code
UX & Visual Polish
4/10
Look and feel quality
Clear visual polish - thoughtful tokens, animations and layered backgrounds improve perceived quality; UX states (loading/error/edge) and undo/optimistic patterns are not present.
Verified artifacts
Expertise
Modern Web Frameworks• Junior
Technologies
JavaScript
TypeScript
Node JS• Junior • 4y+
GraphQL
Next.js
Express
Vercel
Angular
React.js• 4y+
tRPC
Recommendations
- Assign to build polished marketing sites, personal portfolios, and brochure-style product pages using Next.js where strong CSS and responsive layout skills are primary.
- Work on component libraries and design-token refactors - consolidate globals.css into a small design-system and document token usage.
- Contribute UI-focused features in a Next.js frontend (component UI, accessibility improvements, responsive refinements) while pairing with a more senior engineer for state and data-flow architecture.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: