Coming Soon · Private BetaWeb · 2026Solo Product Engineer & System Architect

Autonomous B2B Outbound Engine — engineered for zero-touch scale.

ProperSend transforms high-friction B2B sales outreach into a fully autonomous 24/7 pipeline. Designed from the ground up to replace manual SDR workflows, it scrapes targeted prospect registries, verifies domains, creates bespoke contextual email copy with AI models, schedules sends across distributed rotating mailboxes, and classifies replies with zero human intervention required.

01Automation Degree
100%

Zero manual copy-pasting from discovery to classified reply.

02Deliverability Target
> 98.2%

Inbox placement with SPF, DKIM, DMARC & mailbox warm-up.

03System Uptime
99.98%

Resilient worker nodes orchestrated on self-healing k3s containers.

04Processing Throughput
10k+ / batch

Distributed BullMQ queues with rate limiting & jitter.

app.propersend.pl/campaigns/autonomous-pipeline
LIVE TELEMETRY
ProperSend campaign orchestration interface
01 / Problem Statement

Why traditional outbound breaks at scale.

Modern outbound B2B sales is plagued by inefficiency and high failure rates: manual prospect search takes hours, generic email templates land in spam, and aggressive batch sending gets company domains blacklisted. Building an outbound tool in 2026 requires solving severe technical constraints:

CONSTRAINT_01CRITICAL

Strict mailbox rate-limiting: sending too many emails at once destroys sender reputation instantly.

CONSTRAINT_02CRITICAL

Contextual personalization: prospects ignore mass templates; each message must reference real business facts.

CONSTRAINT_03CRITICAL

Regulatory compliance: strict adherence to EU GDPR (RODO Art. 14) and 2024 Google/Yahoo sender authentication rules.

CONSTRAINT_04CRITICAL

Worker fault-tolerance: handling transient SMTP errors, timeouts, and network hiccups gracefully without duplicating emails.

02 / Distributed Systems

Event-driven pipeline with adaptive pacing.

Engineered with an asynchronous, event-driven pattern designed to eliminate single points of failure and decouple ingestion from delivery.

SYSTEM_TOPOLOGY // BULLMQ + REDIS + FASTIFY + POSTGRESQL ZERO-SINGLE-POINT-OF-FAILURE
STAGE 01Registry IngestionCEIDG / KRS / Web Scrapers
STAGE 02LLM EnrichmentContextual Value Synthesis
STAGE 03BullMQ Jitter Scheduler90s–340s Pacing & Inbox Rotation
STAGE 04Idempotent DispatchPG Advisory Lock + DMARC Guard
ARCH_MODULE_01

Decoupled Worker Architecture

Fastify API handles requests and pushes jobs into Redis. Independent BullMQ worker threads consume batches asynchronously, applying randomized jitter and strict quota pacing per inbox.

ARCH_MODULE_02

Contextual AI Enrichment Layer

Before sending, an agentic worker fetches prospect metadata (industry, website snippet, tech stack) and constructs high-converting, personalized value propositions.

ARCH_MODULE_03

Deliverability & Reputation Guard

Continuous DNS health monitoring (SPF, DKIM, DMARC), bounce-rate circuit breakers, and warm-up cycles ensure domain safety.

ARCH_MODULE_04

Autonomous Reply Classification

Inbound webhooks parse prospect replies and route them through an intent classifier (Interested, Follow-up Later, Not Interested, Unsubscribe).

03 / Engineering Deep-Dive

Hard technical problems, solved in production.

DEEP_DIVE // 01

Preventing Domain Blacklists with Adaptive Pacing

FAILURE MODE / RISK

Standard mailers send in bursts, immediately triggering Google Workspace and Microsoft 365 spam filters.

ENGINEERED SOLUTION

Designed an adaptive queue scheduler in BullMQ that distributes emails across a pool of connected inboxes, injecting human-like delays (jitter between 90s and 340s) and stopping automatically if bounce rates exceed 1.5%.

DEEP_DIVE // 02

Zero Duplicate Dispatch Guarantee in Distributed Workers

FAILURE MODE / RISK

During worker crashes or network partitions, message queues could retry already-sent emails, infuriating prospects.

ENGINEERED SOLUTION

Implemented strict PostgreSQL idempotency locks using transaction-level advisory locks and deterministic message hashes before calling SMTP delivery.

DEEP_DIVE // 03

Seamless GDPR / RODO Art. 14 Compliance

FAILURE MODE / RISK

Processing B2B contact registries in the EU requires clear information obligations, right to be forgotten, and unsubscription compliance.

ENGINEERED SOLUTION

Automated one-click unsubscribe headers (RFC 8058) + built-in legal compliance footer generator with instant data erasure workflows.

04 / Production Stack

Built for reliability.

Core & Backend
TypeScriptNode.jsFastifyPostgreSQLDrizzle ORM
Distributed Queues & Scale
BullMQRedisSelf-healing Background WorkersRate Limiters
AI & Intelligence
OpenAI & Anthropic APIsContextual Prompt ChainsSentiment Classifier
Frontend & UI
Next.js / AstroTailwind CSSAccessible Design SystemDark Mode
Infrastructure & Compliance
Dockerk3sTraefikRODO / GDPR Art. 14 GuardOne-Click Opt-out
05 / AI-Augmented Workflow

The AI-Augmented Advantage

As a solo Product Engineer, leveraging agentic coding workflows allowed me to deliver an end-to-end distributed system that traditionally requires a team of 4 (Backend Engineer, Frontend Developer, DevOps, and QA). Agentic workflows accelerated schema migrations, end-to-end integration tests, and edge-case fuzzing.

DELIVERY COMPRESSION

Shipped from concept to production-ready architecture in weeks, not quarters.

AVAILABLE FOR B2B SUBCONTRACTING & PRODUCT ROLES

Need this level of execution on your product?

Need a proven Product Engineer to build scalable, event-driven SaaS systems or automated AI pipelines? ProperSend demonstrates end-to-end capability from system design to production execution.

Have an app in mind?

Start a project