Gustavo Gaviria Ivanov
Offer Gustavo work on your next project.
Private message
Offer work
Medelln,
Colombia
4 months 25 days back
available for hire
age 31 years
on the service 9 months 9 days
Rating
Successful projects
Average rating
CV
Connecting a Telegram bot to Dify / Web Site
Deploy Ollama with Gemma or Qwen models
Skills and abilities
Portfolio
-
799 USD Bitcoin ECDSA Signature Analysis + Polynonce Scanner
Cryptocurrency & Blockchain
I deliver ready-to-use CLI tools for public Bitcoin data analysis: addresses, transactions, signature parameters (r/s/z), and nonce anomaly detection. CSV exports included.
What’s included (3 scripts):
… S1. HEX → 6 address types + TX/balance check
Automatic address generation from HEX (your test keys) and activity/balance lookup.
Supported variants:
P2PKH (uncompressed)
P2PKH (compressed)
P2SH-P2WPKH (uncompressed)
P2SH-P2WPKH (compressed)
Bech32 P2WPKH (uncompressed)
Bech32 P2WPKH (compressed)
S2. ECDSA (r, s, z) extractor → CSV
Given address/txid, I parse raw transactions (Legacy/SegWit) and export:
txid, input_index, r, s, z, pubkey, prev_tx, prev_index, address_type, hash_type
S3. Polynonce scanner (strict research mode)
Detects nonce anomalies from r/s/z with consistency checks (optionally d·G==Pubkey if available).
Deliverables: CSV files + process logs. Short report on request.
Pricing:
S1 — $129
S2 — $279
S3 — $590
Bundle (S1+S2+S3) — $799
Disclaimer: Tools are for educational and research purposes only. They operate on public blockchain data. No unauthorized actions and no attempts to compromise private keys. The client is responsible for compliance with applicable laws and platform rules.
DM me for sample outputs, demo logs, or minor adjustments to your data format.
-
AEON-Δ: Cognitive Architecture for Autonomous Thought(2025–2026)
AI & Machine LearningAEON‑Δ is a cognitive engine built to simulate full-cycle reasoning — from perception and deep reflection to safe action generation. It operates over abstract latent vectors (z ∈ ℝ²⁵⁶) and models internal cognition through self-reflective loops, memory recall, quantum-style analysis, and safety-aware planning. The system is fully modular, mathematically grounded, and implemented in PyTorch.
Core Modules
… AEONDelta – system orchestrator; controls flow, reasoning, memory, and outputs.
ThoughtEncoder / Decoder – encodes input into latent "thought" vectors and reconstructs text from them.
QualiaExtractor – transforms latent input z into a raw representation ψ₀, initiating the reasoning cycle.
MetaLoopProcessor – core reflective loop computing a fixed point C*, simulating deep contemplation.
PillarsModule – breaks C* into 5 core semantic pillars (interpretable thought axes).
QuantumSimulator – applies unitary operators to compute thought entanglement and internal complexity.
TopologyAnalyzer – analyzes the stability and phase shifts in mental landscapes via Hessians & catastrophe theory.
ActionModule – derives action vector from cognition and evaluates its safety.
PlanningModule – constructs multi-step action plans from current context and predicted outcomes.
RSSM – latent dynamics predictor, enabling temporal continuity of internal thought.
Memory & Knowledge
MemoryManager – episodic/semantic memory handler with vector search and NumPy fallback.
KnowledgeGraph – symbolic SRO-graph knowledge, integrated with Neo4j for scalable querying.
Training Pipeline (Three Phases)
Phase A – ThoughtAETrainer: Trains the autoencoder via curriculum learning and InfoNCE loss; LoRA-adaptable.
Phase B – ZDynamicsTrainer: Learns temporal transitions in latent space (z → z′), building a world model.
Phase C – CuriosityPPOTrainer: Intrinsic-reward based self-play. Agent explores its own latent space, optimizing for novelty, stability, and entropy balance.
Engineering & Innovations
End-to-End Integration – all modules operate synchronously: perception → reasoning → planning.
LoRA support enables fast fine-tuning with low compute overhead.
Safety Assessment – ActionModule filters unsafe or unstable decisions.
Quantum + Topological Reasoning – unique use of entanglement and catastrophe prediction in thought modeling.
Knowledge Injection – symbolic-graph overlay on vector reasoning.
Tools – full save/load cycle, built-in unit tests, CLI inference mode.
AEON‑Δ is a thought engine:
a reasoning system with memory, introspection, planning, and internal reward.
It is built not to imitate text — but to think autonomously.
-
Applied Cryptanalysis & Side-Channel Automation (2024–2025)
ConsultingI specialize in **applied cryptanalytic engineering**, focusing on *low-bit entropy leakage* and *algebraic decoding* from digital signature schemes — primarily in constrained elliptic curve environments (e.g., secp256k1-like groups).
Our workflow integrates strict modular arithmetic control (no heuristics), high-performance lattice reduction (BKZ, LLL), and cohort-based bit slicing.
Tasks performed:
… - Reconstructed private scalars `d` under **LSB-only leak assumptions** with adaptive candidate sweeping.
- Modeled **cohort-variance in Z/r distributions** under multiple nonce policies, using dual-side leakage (MSB+LSB).
- Implemented **multi-phase validators** (oracle-verified) against known public key slices.
- Built full-stack automation pipelines (Python/Sage + fpylll + CSV parsers) to extract viable candidates under bounded Hamming weight.
Additional work includes:
- Side-channel automation with entropy-detecting filters for suspected weak-nonce encodings.
- Structural audits of large-scale signing patterns across several public datasets.
Activity
| Latest proposals 10 | Budget | Added | Deadlines | Proposal | |
|---|---|---|---|---|---|
|
Creation of a TG bot for transmitting market orders to the exchange via API
233 USD
|
|||||
|
Expert in Data Analysis and Web Scraping for Real Estate Market Research in Bogotá
500 USD
|
|||||
|
CRM
15 000 USD
|
|||||
|
Automated content system for Telegram channels based on Make + Google Sheets + OpenAI API.
350 USD
|
|||||
|
Python script for personal recommendations
137 USD
|
|||||
|
Telegram bot with subscription and access to a closed group
116 USD
|
|||||
|
Streamlined, API-based scraping service
250 USD
|
|||||
|
Data parsing bullionbypost
1745 USD
|
|||||
|
Parsing text PDF with tables
137 USD
|
|||||
|
Create a GPT assistant for texts
75 USD
|
