Skip to main content
Rakesh KhadkaBackend, Data & AI Systems

Available for Senior & Staff roles

  • Rabtik Limited
  • Ex-Azminds Services
  • Ex-Aayulogic

I engineerbackends, data & AIthat hold up.

Senior Backend Engineer building Python microservices, AI-ready RAG data pipelines and real-time marketplace sync engines — production systems at RestroExpress, Mobileriz, DatafuseAI and RealHRsoft. Turning messy enterprise data into infrastructure that stays up.

Explore deployed case studiesView company experienceSchedule a technical consultation

Multi-engine ETL

50+sources

DatafuseAI Spark, Databricks & Livy

Marketplace gateway

3unified

Trendyol, Hepsiburada, Çiçeksepeti

Enterprise scale

50k+users

RealHRsoft monthly active payroll

Core reliability

99.99%SLA

Production multi-tenant workloads

  • Python & Django
  • FastAPI
  • AsyncIO
  • vLLM & LangChain
  • pgvector & Qdrant
  • PostgreSQL & Redis
  • Celery workers
  • Apache Kafka
  • ClickHouse
  • AWS ECS / EKS
  • Docker & CI/CD
  • WebSockets

Engineering focus

Core architecture pillars

Deep expertise combining enterprise-grade backend reliability with modern vector retrieval pipelines and real-time distributed data layers.

Technical toolkit

Engineered stack matrix

Structured engineering arsenal organised by domain and battle-tested across DatafuseAI, RealHRsoft and Mobileriz.

Production deployments

Featured architecture case studies

Real-world systems delivering measurable business impact across payroll processing, enterprise RAG pipelines and real-time marketplace sync.

RealHRsoft // 01Enterprise HRMS

RealHR payroll & workforce engine

Engineered the multi-tenant payroll calculation engine and real-time attendance processing microservice for a high-scale enterprise HRMS serving tens of thousands of corporate employees. Handled complex tax rule engines, overtime calculation, Celery background worker pipelines and third-party integrations with zero margin of error.

Active employees
50k+
Payroll runtime
−65%
Calc accuracy
100.0%
  • Python
  • Django & DRF
  • PostgreSQL
  • Celery & Redis
  • AWS ECS

Workforce payroll worker

RealHR payroll processing architectureAttendance events feed a Celery worker running tax and salary rules, which writes results to PostgreSQL and dispatches notifications to Stripe and Slack.ATTENDANCECELERY WORKERTax & salary rulesPOSTGRES DBSTRIPE / SLACK

Batch: 50,000 payslipsProcessed in 3.4m

DatafuseAI // 02AI & RAG data preparation

Enterprise AI & RAG data preparation engine

Architected enterprise pipelines standardising, deduplicating and profiling massive multi-source datasets — 50+ sources including Snowflake, Postgres and BigQuery — into AI-ready vector formats. Fed clean, validated chunk embeddings to pgvector and Qdrant under strict schema guardrails, eliminating hallucination risk and enabling governed context retrieval.

Source ingestion
50+sources
Vector targets
pgvector / Qdrant
Clean data pass
99.8%
  • Qdrant & pgvector
  • Apache Spark
  • Apache Livy
  • Deduplication engine
  • Context guardrails

RAG ingestion & vector pipeline

Deduplication active

RAG ingestion and vector pipeline architectureFifty-plus enterprise sources and raw datasets flow into a Spark deduplication and schema validation stage, which writes clean embeddings to a Qdrant or pgvector store.50+ SOURCESRAW DATASETSSPARK + DEDUPValidation & schemaVECTOR DBQdrant/pgvector

Zero-hallucination embeddings50+ connectors

Mobileriz // 03E-commerce & AI sync

Multi-marketplace sync & AI operations hub

High-concurrency seller middleware synchronising Trendyol, Hepsiburada and Çiçeksepeti with embedded AI workflows: automated triage and draft responses for customer support, semantic embeddings for unified cross-marketplace catalogue mapping, and real-time anomaly detection flagging return fraud.

AI support triage
Unified inbox
Attribute mapping
LLM embeddings
  • Trendyol API
  • Hepsiburada API
  • Çiçeksepeti API
  • AI anomaly triage

Infrastructure // 04Distributed ETL

Data streaming & ETL orchestration

Fault-tolerant streaming ingestion mesh built on Apache Kafka and ClickHouse, powering sub-second aggregation for audit logs, real-time analytics and automated telemetry alerts across multi-zone container environments.

Throughput
120k/s
End-to-end lag
<80ms
  • Kafka
  • ClickHouse
  • Docker
  • AWS ECS

Career trajectory

Professional experience & impact

A track record of architecting core backends, intelligent data pipelines and resilient distributed infrastructure.

  1. Software Engineer

    Rabtik Limited

    RestroExpress

    Current

    Working across both application and infrastructure — building product features while standing up and maintaining the environments they run on.

    • Set up and operate the production infrastructure: environment provisioning, deployment pipelines and release automation.
    • Own features across the full lifecycle — requirement breakdown and system design, implementation, code review, deployment and post-release monitoring.

  2. Backend & Integrations Engineer

    Azminds Services

    Mobileriz

    Engineered core e-commerce marketplace synchronisation infrastructure and intelligent automation connecting multi-store merchants with Trendyol, Hepsiburada and Çiçeksepeti.

    • Engineered bidirectional marketplace connector APIs ensuring sub-second inventory, catalogue and price synchronisation.
    • Developed AI-powered support message triage and draft response generation across a unified multi-marketplace inbox.
    • Implemented embedding-based category and attribute mapping across divergent marketplace schemas, plus automated return-fraud anomaly detection.
    • Architected secure multi-tenant token vaults, 2FA and high-throughput webhook dispatch pipelines with Redis.
    • Python & FastAPI
    • AI support triage
    • Semantic attribute mapping
    • Anomaly detection
    • Trendyol API
    • Hepsiburada API
    • Çiçeksepeti API

  3. Senior Data Systems & AI Infrastructure Engineer

    Azminds Services

    DatafuseAI

    Spearheaded multi-engine data preparation pipelines, deduplication frameworks and automated schema validation feeding vector databases and RAG pipelines for enterprise LLMs.

    • Architected high-throughput preparation and chunking pipelines standardising and deduplicating 50+ enterprise sources (Snowflake, BigQuery, Postgres, S3) into clean high-dimensional embeddings.
    • Built automated schema profiling, entity resolution and data-quality validation that prevents model hallucination and guarantees clean context retrieval downstream.
    • Optimised distributed transformation and indexing jobs on Apache Spark, Databricks and Livy, streaming pre-processed datasets straight into pgvector and Qdrant clusters.
    • Apache Spark
    • Qdrant & pgvector
    • Apache Livy
    • AI data prep
    • Semantic deduplication
    • Context guardrails

  4. Software Engineer (Backend & Infrastructure)

    Aayulogic

    RealHRsoft

    Engineered the core backend microservices and computational engine powering a high-scale enterprise Human Resource Management System.

    • Developed and scaled the multi-tenant payroll calculation engine and real-time biometric attendance sync pipelines serving 50,000+ corporate employees.
    • Streamlined PostgreSQL query execution, index architecture and Celery/Redis distributed workers, cutting payroll processing cycle times by 65%.
    • Built third-party enterprise integrations including Stripe for automated payments and Slack for workforce notifications.
    • Automated CI/CD pipelines and AWS container infrastructure with Docker and GitHub Actions, sustaining 99.98% service uptime.
    • Python
    • Django & DRF
    • PostgreSQL
    • Celery
    • Redis
    • AWS ECS
    • Docker
    • GitHub Actions

Benchmark telemetry

Measured performance benchmarks

Concrete engineering gains delivered across DatafuseAI, RealHRsoft and Mobileriz systems.

RAG data prep pass rate

99.8%

DatafuseAI schema validation, deduplication & vector embeddings

Payroll runtime

−65%

RealHRsoft Celery & PostgreSQL indexing refactor

Support triage time

−70%

Mobileriz AI auto-triage & marketplace sync routing

System availability

99.99%

Maintained across production multi-tenant workloads

Initiate contact · system ready

Have a project? Let's build together.

Open to Senior/Staff Software Engineer roles, high-scale backend consulting and AI/RAG system architecture. Guaranteed response within 24 hours.

npx contact-engineer --role senior

Email me directly

Contact form